Datasets:

bunkalab

/

arxiv_datasets_llm

like 0

Causal Graph Discovery with Retrieval-Augmented Generation based Large Language Models

Yuzhe Zhang, Yipeng Zhang, Yidong Gan, Lina Yao, Chen Wang

Causal graph recovery is essential in the field of causal inference. Traditional methods are typically knowledge-based or statistical estimation-based, which are limited by data collection biases and individuals' knowledge about factors affecting the relations between variables of interests. The advance of large language models (LLMs) provides opportunities to address these problems. We propose a novel method that utilizes the extensive knowledge contained within a large corpus of scientific literature to deduce causal relationships in general causal graph recovery tasks. This method leverages Retrieval Augmented-Generation (RAG) based LLMs to systematically analyze and extract pertinent information from a comprehensive collection of research papers. Our method first retrieves relevant text chunks from the aggregated literature. Then, the LLM is tasked with identifying and labelling potential associations between factors. Finally, we give a method to aggregate the associational relationships to build a causal graph. We demonstrate our method is able to construct high quality causal graphs on the well-known SACHS dataset solely from literature.

http://arxiv.org/abs/2402.15301v1

cs.CL, cs.LG, stat.ME

GPT-HateCheck: Can LLMs Write Better Functional Tests for Hate Speech Detection?

Yiping Jin, Leo Wanner, Alexander Shvets

Online hate detection suffers from biases incurred in data sampling, annotation, and model pre-training. Therefore, measuring the averaged performance over all examples in held-out test data is inadequate. Instead, we must identify specific model weaknesses and be informed when it is more likely to fail. A recent proposal in this direction is HateCheck, a suite for testing fine-grained model functionalities on synthesized data generated using templates of the kind "You are just a [slur] to me." However, despite enabling more detailed diagnostic insights, the HateCheck test cases are often generic and have simplistic sentence structures that do not match the real-world data. To address this limitation, we propose GPT-HateCheck, a framework to generate more diverse and realistic functional tests from scratch by instructing large language models (LLMs). We employ an additional natural language inference (NLI) model to verify the generations. Crowd-sourced annotation demonstrates that the generated test cases are of high quality. Using the new functional tests, we can uncover model weaknesses that would be overlooked using the original HateCheck dataset.

http://arxiv.org/abs/2402.15238v1

cs.CL, cs.CY

Item-side Fairness of Large Language Model-based Recommendation System

Meng Jiang, Keqin Bao, Jizhi Zhang, Wenjie Wang, Zhengyi Yang, Fuli Feng, Xiangnan He

Recommendation systems for Web content distribution intricately connect to the information access and exposure opportunities for vulnerable populations. The emergence of Large Language Models-based Recommendation System (LRS) may introduce additional societal challenges to recommendation systems due to the inherent biases in Large Language Models (LLMs). From the perspective of item-side fairness, there remains a lack of comprehensive investigation into the item-side fairness of LRS given the unique characteristics of LRS compared to conventional recommendation systems. To bridge this gap, this study examines the property of LRS with respect to item-side fairness and reveals the influencing factors of both historical users' interactions and inherent semantic biases of LLMs, shedding light on the need to extend conventional item-side fairness methods for LRS. Towards this goal, we develop a concise and effective framework called IFairLRS to enhance the item-side fairness of an LRS. IFairLRS covers the main stages of building an LRS with specifically adapted strategies to calibrate the recommendations of LRS. We utilize IFairLRS to fine-tune LLaMA, a representative LLM, on \textit{MovieLens} and \textit{Steam} datasets, and observe significant item-side fairness improvements. The code can be found in https://github.com/JiangM-C/IFairLRS.git.

http://arxiv.org/abs/2402.15215v1

cs.IR

GraphEdit: Large Language Models for Graph Structure Learning

Zirui Guo, Lianghao Xia, Yanhua Yu, Yuling Wang, Zixuan Yang, Wei Wei, Liang Pang, Tat-Seng Chua, Chao Huang

Graph Structure Learning (GSL) focuses on capturing intrinsic dependencies and interactions among nodes in graph-structured data by generating novel graph structures. Graph Neural Networks (GNNs) have emerged as promising GSL solutions, utilizing recursive message passing to encode node-wise inter-dependencies. However, many existing GSL methods heavily depend on explicit graph structural information as supervision signals, leaving them susceptible to challenges such as data noise and sparsity. In this work, we propose GraphEdit, an approach that leverages large language models (LLMs) to learn complex node relationships in graph-structured data. By enhancing the reasoning capabilities of LLMs through instruction-tuning over graph structures, we aim to overcome the limitations associated with explicit graph structural information and enhance the reliability of graph structure learning. Our approach not only effectively denoises noisy connections but also identifies node-wise dependencies from a global perspective, providing a comprehensive understanding of the graph structure. We conduct extensive experiments on multiple benchmark datasets to demonstrate the effectiveness and robustness of GraphEdit across various settings. We have made our model implementation available at: https://github.com/HKUDS/GraphEdit.

http://arxiv.org/abs/2402.15183v4

cs.LG, cs.AI

Machine Unlearning of Pre-trained Large Language Models

Jin Yao, Eli Chien, Minxin Du, Xinyao Niu, Tianhao Wang, Zezhou Cheng, Xiang Yue

This study investigates the concept of the `right to be forgotten' within the context of large language models (LLMs). We explore machine unlearning as a pivotal solution, with a focus on pre-trained models--a notably under-researched area. Our research delineates a comprehensive framework for machine unlearning in pre-trained LLMs, encompassing a critical analysis of seven diverse unlearning methods. Through rigorous evaluation using curated datasets from arXiv, books, and GitHub, we establish a robust benchmark for unlearning performance, demonstrating that these methods are over $10^5$ times more computationally efficient than retraining. Our results show that integrating gradient ascent with gradient descent on in-distribution data improves hyperparameter robustness. We also provide detailed guidelines for efficient hyperparameter tuning in the unlearning process. Our findings advance the discourse on ethical AI practices, offering substantive insights into the mechanics of machine unlearning for pre-trained LLMs and underscoring the potential for responsible AI development.

http://arxiv.org/abs/2402.15159v2

cs.CL, cs.AI, cs.CR, cs.LG

Where Visual Speech Meets Language: VSP-LLM Framework for Efficient and Context-Aware Visual Speech Processing

Jeong Hun Yeo, Seunghee Han, Minsu Kim, Yong Man Ro

In visual speech processing, context modeling capability is one of the most important requirements due to the ambiguous nature of lip movements. For example, homophenes, words that share identical lip movements but produce different sounds, can be distinguished by considering the context. In this paper, we propose a novel framework, namely Visual Speech Processing incorporated with LLMs (VSP-LLM), to maximize the context modeling ability by bringing the overwhelming power of LLMs. Specifically, VSP-LLM is designed to perform multi-tasks of visual speech recognition and translation, where the given instructions control the type of task. The input video is mapped to the input latent space of a LLM by employing a self-supervised visual speech model. Focused on the fact that there is redundant information in input frames, we propose a novel deduplication method that reduces the embedded visual features by employing visual speech units. Through the proposed deduplication and Low Rank Adaptors (LoRA), VSP-LLM can be trained in a computationally efficient manner. In the translation dataset, the MuAViC benchmark, we demonstrate that VSP-LLM can more effectively recognize and translate lip movements with just 15 hours of labeled data, compared to the recent translation model trained with 433 hours of labeld data.

http://arxiv.org/abs/2402.15151v1

cs.CV, cs.CL, eess.AS, eess.IV

Improving Sentence Embeddings with an Automatically Generated NLI Dataset

Soma Sato, Hayato Tsukagoshi, Ryohei Sasano, Koichi Takeda

Decoder-based large language models (LLMs) have shown high performance on many tasks in natural language processing. This is also true for sentence embedding learning, where a decoder-based model, PromptEOL, has achieved the best performance on semantic textual similarity (STS) tasks. However, PromptEOL makes great use of fine-tuning with a manually annotated natural language inference (NLI) dataset. We aim to improve sentence embeddings learned in an unsupervised setting by automatically generating an NLI dataset with an LLM and using it to fine-tune PromptEOL. In experiments on STS tasks, the proposed method achieved an average Spearman's rank correlation coefficient of 82.21 with respect to human evaluation, thus outperforming existing methods without using large, manually annotated datasets.

http://arxiv.org/abs/2402.15132v1

cs.CL, cs.LG

Interactive-KBQA: Multi-Turn Interactions for Knowledge Base Question Answering with Large Language Models

Guanming Xiong, Junwei Bao, Wen Zhao

This study explores the realm of knowledge-base question answering (KBQA). KBQA is considered a challenging task, particularly in parsing intricate questions into executable logical forms. Traditional semantic parsing (SP)-based methods require extensive data annotations, which result in significant costs. Recently, the advent of few-shot in-context learning, powered by large language models (LLMs), has showcased promising capabilities. Yet, fully leveraging LLMs to parse questions into logical forms in low-resource scenarios poses a substantial challenge. To tackle these hurdles, we introduce Interactive-KBQA, a framework designed to generate logical forms through direct interaction with knowledge bases (KBs). Within this framework, we have developed three generic APIs for KB interaction. For each category of complex question, we devised exemplars to guide LLMs through the reasoning processes. Our method achieves competitive results on the WebQuestionsSP, ComplexWebQuestions, KQA Pro, and MetaQA datasets with a minimal number of examples (shots). Importantly, our approach supports manual intervention, allowing for the iterative refinement of LLM outputs. By annotating a dataset with step-wise reasoning processes, we showcase our model's adaptability and highlight its potential for contributing significant enhancements to the field.

http://arxiv.org/abs/2402.15131v1

cs.CL, cs.AI, I.2.7

AttributionBench: How Hard is Automatic Attribution Evaluation?

Yifei Li, Xiang Yue, Zeyi Liao, Huan Sun

Modern generative search engines enhance the reliability of large language model (LLM) responses by providing cited evidence. However, evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem. This verification, traditionally dependent on costly human evaluation, underscores the urgent need for automatic attribution evaluation methods. To bridge the gap in the absence of standardized benchmarks for these methods, we present AttributionBench, a comprehensive benchmark compiled from various existing attribution datasets. Our extensive experiments on AttributionBench reveal the challenges of automatic attribution evaluation, even for state-of-the-art LLMs. Specifically, our findings show that even a fine-tuned GPT-3.5 only achieves around 80% macro-F1 under a binary classification formulation. A detailed analysis of more than 300 error cases indicates that a majority of failures stem from the model's inability to process nuanced information, and the discrepancy between the information the model has access to and that human annotators do.

http://arxiv.org/abs/2402.15089v1

cs.CL, cs.AI, cs.LG

Getting Serious about Humor: Crafting Humor Datasets with Unfunny Large Language Models

Zachary Horvitz, Jingru Chen, Rahul Aditya, Harshvardhan Srivastava, Robert West, Zhou Yu, Kathleen McKeown

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. In our work, 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.

http://arxiv.org/abs/2403.00794v1

cs.CL, cs.AI, cs.LG

Gotcha! Don't trick me with unanswerable questions! Self-aligning Large Language Models for Responding to Unknown Questions

Yang Deng, Yong Zhao, Moxin Li, See-Kiong Ng, Tat-Seng Chua

Despite the remarkable abilities of Large Language Models (LLMs) to answer questions, they often display a considerable level of overconfidence even when the question does not have a definitive answer. To avoid providing hallucinated answers to these unknown questions, existing studies typically investigate approaches to refusing to answer these questions. In this work, we propose a novel and scalable self-alignment method to utilize the LLM itself to enhance its response-ability to different types of unknown questions, being capable of not only refusing to answer but also providing explanation to the unanswerability of unknown questions. Specifically, the Self-Align method first employ a two-stage class-aware self-augmentation approach to generate a large amount of unknown question-response data. Then we conduct disparity-driven self-curation to select qualified data for fine-tuning the LLM itself for aligning the responses to unknown questions as desired. Experimental results on two datasets across four types of unknown questions validate the superiority of the Self-Align method over existing baselines in terms of three types of task formulation.

http://arxiv.org/abs/2402.15062v1

cs.CL, cs.LG

Fine-tuning Large Language Models for Domain-specific Machine Translation

Jiawei Zheng, Hanghai Hong, Xiaoli Wang, Jingsong Su, Yonggui Liang, Shikai Wu

Large language models (LLMs) have made significant progress in machine translation (MT). However, their potential in domain-specific MT remains under-explored. Current LLM-based MT systems still face several challenges. First, for LLMs with in-context learning, their effectiveness is highly sensitive to input translation examples, and processing them can increase inference costs. They often require extra post-processing due to over-generation. Second, LLMs with fine-tuning on domain-specific data often require high training costs for domain adaptation, and may weaken the zero-shot MT capabilities of LLMs due to over-specialization. The aforementioned methods can struggle to translate rare words in domain transfer scenarios. To address these challenges, this paper proposes a prompt-oriented fine-tuning method, denoted as LlamaIT, to effectively and efficiently fine-tune a general-purpose LLM for domain-specific MT tasks. First, we construct a task-specific mix-domain dataset, which is then used to fine-tune the LLM with LoRA. This can eliminate the need for input translation examples, post-processing, or over-specialization. By zero-shot prompting with instructions, we adapt the MT tasks to the target domain at inference time. To further elicit the MT capability for rare words, we construct new prompts by incorporating domain-specific bilingual vocabulary. We also conduct extensive experiments on both publicly available and self-constructed datasets. The results show that our LlamaIT can significantly enhance the domain-specific MT capabilities of the LLM, meanwhile preserving its zero-shot MT capabilities.

http://arxiv.org/abs/2402.15061v1

cs.CL, cs.LG

On the Multi-turn Instruction Following for Conversational Web Agents

Yang Deng, Xuan Zhang, Wenxuan Zhang, Yifei Yuan, See-Kiong Ng, Tat-Seng Chua

Web agents powered by Large Language Models (LLMs) have demonstrated remarkable abilities in planning and executing multi-step interactions within complex web-based environments, fulfilling a wide range of web navigation tasks. Despite these advancements, the potential for LLM-powered agents to effectively engage with sequential user instructions in real-world scenarios has not been fully explored. In this work, we introduce a new task of Conversational Web Navigation, which necessitates sophisticated interactions that span multiple turns with both the users and the environment, supported by a specially developed dataset named Multi-Turn Mind2Web (MT-Mind2Web). To tackle the limited context length of LLMs and the context-dependency issue of the conversational tasks, we further propose a novel framework, named self-reflective memory-augmented planning (Self-MAP), which employs memory utilization and self-reflection techniques. Extensive experiments are conducted to benchmark the MT-Mind2Web dataset, and validate the effectiveness of the proposed method.

http://arxiv.org/abs/2402.15057v1

cs.CL, cs.AI

KIEval: A Knowledge-grounded Interactive Evaluation Framework for Large Language Models

Zhuohao Yu, Chang Gao, Wenjin Yao, Yidong Wang, Wei Ye, Jindong Wang, Xing Xie, Yue Zhang, Shikun Zhang

Automatic evaluation methods for large language models (LLMs) are hindered by data contamination, leading to inflated assessments of their effectiveness. Existing strategies, which aim to detect contaminated texts, focus on quantifying contamination status instead of accurately gauging model performance. In this paper, we introduce KIEval, a Knowledge-grounded Interactive Evaluation framework, which incorporates an LLM-powered "interactor" role for the first time to accomplish a dynamic contamination-resilient evaluation. Starting with a question in a conventional LLM benchmark involving domain-specific knowledge, KIEval utilizes dynamically generated, multi-round, and knowledge-focused dialogues to determine whether a model's response is merely a recall of benchmark answers or demonstrates a deep comprehension to apply knowledge in more complex conversations. Extensive experiments on seven leading LLMs across five datasets validate KIEval's effectiveness and generalization. We also reveal that data contamination brings no contribution or even negative effect to models' real-world applicability and understanding, and existing contamination detection methods for LLMs can only identify contamination in pre-training but not during supervised fine-tuning.

http://arxiv.org/abs/2402.15043v1

cs.CL, cs.AI, cs.LG

Divide-or-Conquer? Which Part Should You Distill Your LLM?

Zhuofeng Wu, He Bai, Aonan Zhang, Jiatao Gu, VG Vinod Vydiswaran, Navdeep Jaitly, Yizhe Zhang

Recent methods have demonstrated that Large Language Models (LLMs) can solve reasoning tasks better when they are encouraged to solve subtasks of the main task first. In this paper we devise a similar strategy that breaks down reasoning tasks into a problem decomposition phase and a problem solving phase and show that the strategy is able to outperform a single stage solution. Further, we hypothesize that the decomposition should be easier to distill into a smaller model compared to the problem solving because the latter requires large amounts of domain knowledge while the former only requires learning general problem solving strategies. We propose methods to distill these two capabilities and evaluate their impact on reasoning outcomes and inference cost. We find that we can distill the problem decomposition phase and at the same time achieve good generalization across tasks, datasets, and models. However, it is harder to distill the problem solving capability without losing performance and the resulting distilled model struggles with generalization. These results indicate that by using smaller, distilled problem decomposition models in combination with problem solving LLMs we can achieve reasoning with cost-efficient inference and local adaptation.

http://arxiv.org/abs/2402.15000v1

cs.CL, cs.LG

Optimizing Language Models for Human Preferences is a Causal Inference Problem

Victoria Lin, Eli Ben-Michael, Louis-Philippe Morency

As large language models (LLMs) see greater use in academic and commercial settings, there is increasing interest in methods that allow language models to generate texts aligned with human preferences. In this paper, we present an initial exploration of language model optimization for human preferences from direct outcome datasets, where each sample consists of a text and an associated numerical outcome measuring the reader's response. We first propose that language model optimization should be viewed as a causal problem to ensure that the model correctly learns the relationship between the text and the outcome. We formalize this causal language optimization problem, and we develop a method--causal preference optimization (CPO)--that solves an unbiased surrogate objective for the problem. We further extend CPO with doubly robust CPO (DR-CPO), which reduces the variance of the surrogate objective while retaining provably strong guarantees on bias. Finally, we empirically demonstrate the effectiveness of (DR-)CPO in optimizing state-of-the-art LLMs for human preferences on direct outcome data, and we validate the robustness of DR-CPO under difficult confounding conditions.

http://arxiv.org/abs/2402.14979v1

cs.LG, cs.CL, stat.ME

MultiLS: A Multi-task Lexical Simplification Framework

Kai North, Tharindu Ranasinghe, Matthew Shardlow, Marcos Zampieri

Lexical Simplification (LS) automatically replaces difficult to read words for easier alternatives while preserving a sentence's original meaning. LS is a precursor to Text Simplification with the aim of improving text accessibility to various target demographics, including children, second language learners, individuals with reading disabilities or low literacy. Several datasets exist for LS. These LS datasets specialize on one or two sub-tasks within the LS pipeline. However, as of this moment, no single LS dataset has been developed that covers all LS sub-tasks. We present MultiLS, the first LS framework that allows for the creation of a multi-task LS dataset. We also present MultiLS-PT, the first dataset to be created using the MultiLS framework. We demonstrate the potential of MultiLS-PT by carrying out all LS sub-tasks of (1). lexical complexity prediction (LCP), (2). substitute generation, and (3). substitute ranking for Portuguese. Model performances are reported, ranging from transformer-based models to more recent large language models (LLMs).

http://arxiv.org/abs/2402.14972v1

cs.CL, cs.AI

Mitigating Fine-tuning Jailbreak Attack with Backdoor Enhanced Alignment

Jiongxiao Wang, Jiazhao Li, Yiquan Li, Xiangyu Qi, Junjie Hu, Yixuan Li, Patrick McDaniel, Muhao Chen, Bo Li, Chaowei Xiao

Despite the general capabilities of Large Language Models (LLMs) like GPT-4 and Llama-2, these models still request fine-tuning or adaptation with customized data when it comes to meeting the specific business demands and intricacies of tailored use cases. However, this process inevitably introduces new safety threats, particularly against the Fine-tuning based Jailbreak Attack (FJAttack), where incorporating just a few harmful examples into the fine-tuning dataset can significantly compromise the model safety. Though potential defenses have been proposed by incorporating safety examples into the fine-tuning dataset to reduce the safety issues, such approaches require incorporating a substantial amount of safety examples, making it inefficient. To effectively defend against the FJAttack with limited safety examples, we propose a Backdoor Enhanced Safety Alignment method inspired by an analogy with the concept of backdoor attacks. In particular, we construct prefixed safety examples by integrating a secret prompt, acting as a "backdoor trigger", that is prefixed to safety examples. Our comprehensive experiments demonstrate that through the Backdoor Enhanced Safety Alignment with adding as few as 11 prefixed safety examples, the maliciously fine-tuned LLMs will achieve similar safety performance as the original aligned models. Furthermore, we also explore the effectiveness of our method in a more practical setting where the fine-tuning data consists of both FJAttack examples and the fine-tuning task data. Our method shows great efficacy in defending against FJAttack without harming the performance of fine-tuning tasks.

http://arxiv.org/abs/2402.14968v2

cs.CR, cs.CL

Mirror: A Multiple-perspective Self-Reflection Method for Knowledge-rich Reasoning

Hanqi Yan, Qinglin Zhu, Xinyu Wang, Lin Gui, Yulan He

While Large language models (LLMs) have the capability to iteratively reflect on their own outputs, recent studies have observed their struggles with knowledge-rich problems without access to external resources. In addition to the inefficiency of LLMs in self-assessment, we also observe that LLMs struggle to revisit their predictions despite receiving explicit negative feedback. Therefore, We propose Mirror, a Multiple-perspective self-reflection method for knowledge-rich reasoning, to avoid getting stuck at a particular reflection iteration. Mirror enables LLMs to reflect from multiple-perspective clues, achieved through a heuristic interaction between a Navigator and a Reasoner. It guides agents toward diverse yet plausibly reliable reasoning trajectory without access to ground truth by encouraging (1) diversity of directions generated by Navigator and (2) agreement among strategically induced perturbations in responses generated by the Reasoner. The experiments on five reasoning datasets demonstrate that Mirror's superiority over several contemporary self-reflection approaches. Additionally, the ablation study studies clearly indicate that our strategies alleviate the aforementioned challenges.

http://arxiv.org/abs/2402.14963v1

cs.CL, cs.AI

CriticBench: Benchmarking LLMs for Critique-Correct Reasoning

Zicheng Lin, Zhibin Gou, Tian Liang, Ruilin Luo, Haowei Liu, Yujiu Yang

The ability of Large Language Models (LLMs) to critique and refine their reasoning is crucial for their application in evaluation, feedback provision, and self-improvement. This paper introduces CriticBench, a comprehensive benchmark designed to assess LLMs' abilities to critique and rectify their reasoning across a variety of tasks. CriticBench encompasses five reasoning domains: mathematical, commonsense, symbolic, coding, and algorithmic. It compiles 15 datasets and incorporates responses from three LLM families. Utilizing CriticBench, we evaluate and dissect the performance of 17 LLMs in generation, critique, and correction reasoning, i.e., GQC reasoning. Our findings reveal: (1) a linear relationship in GQC capabilities, with critique-focused training markedly enhancing performance; (2) a task-dependent variation in correction effectiveness, with logic-oriented tasks being more amenable to correction; (3) GQC knowledge inconsistencies that decrease as model size increases; and (4) an intriguing inter-model critiquing dynamic, where stronger models are better at critiquing weaker ones, while weaker models can surprisingly surpass stronger ones in their self-critique. We hope these insights into the nuanced critique-correct reasoning of LLMs will foster further research in LLM critique and self-improvement.

http://arxiv.org/abs/2402.14809v2

cs.CL, cs.AI, cs.LG

Enhancing Systematic Decompositional Natural Language Inference Using Informal Logic

Nathaniel Weir, Kate Sanders, Orion Weller, Shreya Sharma, Dongwei Jiang, Zhengping Jiang, Bhavana Dalvi Mishra, Oyvind Tafjord, Peter Jansen, Peter Clark, Benjamin Van Durme

Contemporary language models enable new opportunities for structured reasoning with text, such as the construction and evaluation of intuitive, proof-like textual entailment trees without relying on brittle formal logic. However, progress in this direction has been hampered by a long-standing lack of a clear protocol for determining what valid compositional entailment is. This absence causes noisy datasets and limited performance gains by modern neuro-symbolic engines. To address these problems, we formulate a consistent and theoretically grounded approach to annotating decompositional entailment datasets, and evaluate its impact on LLM-based textual inference. We find that our resulting dataset, RDTE (Recognizing Decompositional Textual Entailment), has a substantially higher internal consistency (+9%) than prior decompositional entailment datasets, suggesting that RDTE is a significant step forward in the long-standing problem of forming a clear protocol for discerning entailment. We also find that training an RDTE-oriented entailment classifier via knowledge distillation and employing it in a modern neuro-symbolic reasoning engine significantly improves results (both accuracy and proof quality) over other entailment classifier baselines, illustrating the practical benefit of this advance for textual inference.

http://arxiv.org/abs/2402.14798v2

cs.CL, cs.AI

IEPile: Unearthing Large-Scale Schema-Based Information Extraction Corpus

Honghao Gui, Lin Yuan, Hongbin Ye, Ningyu Zhang, Mengshu Sun, Lei Liang, Huajun Chen

Large Language Models (LLMs) demonstrate remarkable potential across various domains; however, they exhibit a significant performance gap in Information Extraction (IE). Note that high-quality instruction data is the vital key for enhancing the specific capabilities of LLMs, while current IE datasets tend to be small in scale, fragmented, and lack standardized schema. To this end, we introduce IEPile, a comprehensive bilingual (English and Chinese) IE instruction corpus, which contains approximately 0.32B tokens. We construct IEPile by collecting and cleaning 33 existing IE datasets, and introduce schema-based instruction generation to unearth a large-scale corpus. Experimental results on LLaMA, Baichuan and Qwen demonstrate that using IEPile can enhance the performance of LLMs for IE, especially the zero-shot generalization. We open-source the resource and pre-trained models, hoping to provide valuable support to the NLP community.

http://arxiv.org/abs/2402.14710v2

cs.CL, cs.AI, cs.DB, cs.IR, cs.LG

An LLM-Enhanced Adversarial Editing System for Lexical Simplification

Keren Tan, Kangyang Luo, Yunshi Lan, Zheng Yuan, Jinlong Shu

Lexical Simplification (LS) aims to simplify text at the lexical level. Existing methods rely heavily on annotated data, making it challenging to apply in low-resource scenarios. In this paper, we propose a novel LS method without parallel corpora. This method employs an Adversarial Editing System with guidance from a confusion loss and an invariance loss to predict lexical edits in the original sentences. Meanwhile, we introduce an innovative LLM-enhanced loss to enable the distillation of knowledge from Large Language Models (LLMs) into a small-size LS system. From that, complex words within sentences are masked and a Difficulty-aware Filling module is crafted to replace masked positions with simpler words. At last, extensive experimental results and analyses on three benchmark LS datasets demonstrate the effectiveness of our proposed method.

http://arxiv.org/abs/2402.14704v3

cs.CL

InfFeed: Influence Functions as a Feedback to Improve the Performance of Subjective Tasks

Somnath Banerjee, Maulindu Sarkar, Punyajoy Saha, Binny Mathew, Animesh Mukherjee

Recently, influence functions present an apparatus for achieving explainability for deep neural models by quantifying the perturbation of individual train instances that might impact a test prediction. Our objectives in this paper are twofold. First we incorporate influence functions as a feedback into the model to improve its performance. Second, in a dataset extension exercise, using influence functions to automatically identify data points that have been initially `silver' annotated by some existing method and need to be cross-checked (and corrected) by annotators to improve the model performance. To meet these objectives, in this paper, we introduce InfFeed, which uses influence functions to compute the influential instances for a target instance. Toward the first objective, we adjust the label of the target instance based on its influencer(s) label. In doing this, InfFeed outperforms the state-of-the-art baselines (including LLMs) by a maximum macro F1-score margin of almost 4% for hate speech classification, 3.5% for stance classification, and 3% for irony and 2% for sarcasm detection. Toward the second objective we show that manually re-annotating only those silver annotated data points in the extension set that have a negative influence can immensely improve the model performance bringing it very close to the scenario where all the data points in the extension set have gold labels. This allows for huge reduction of the number of data points that need to be manually annotated since out of the silver annotated extension dataset, the influence function scheme picks up ~1/1000 points that need manual correction.

http://arxiv.org/abs/2402.14702v2

cs.CL

COMPASS: Computational Mapping of Patient-Therapist Alliance Strategies with Language Modeling

Baihan Lin, Djallel Bouneffouf, Yulia Landa, Rachel Jespersen, Cheryl Corcoran, Guillermo Cecchi

The therapeutic working alliance is a critical factor in predicting the success of psychotherapy treatment. Traditionally, working alliance assessment relies on questionnaires completed by both therapists and patients. In this paper, we present COMPASS, a novel framework to directly infer the therapeutic working alliance from the natural language used in psychotherapy sessions. Our approach utilizes advanced large language models to analyze transcripts of psychotherapy sessions and compare them with distributed representations of statements in the working alliance inventory. Analyzing a dataset of over 950 sessions covering diverse psychiatric conditions, we demonstrate the effectiveness of our method in microscopically mapping patient-therapist alignment trajectories and providing interpretability for clinical psychiatry and in identifying emerging patterns related to the condition being treated. By employing various neural topic modeling techniques in combination with generative language prompting, we analyze the topical characteristics of different psychiatric conditions and incorporate temporal modeling to capture the evolution of topics at a turn-level resolution. This combined framework enhances the understanding of therapeutic interactions, enabling timely feedback for therapists regarding conversation quality and providing interpretable insights to improve the effectiveness of psychotherapy.

http://arxiv.org/abs/2402.14701v1

cs.CL, cs.AI, cs.HC, cs.LG, q-bio.NC

UFO: a Unified and Flexible Framework for Evaluating Factuality of Large Language Models

Zhaoheng Huang, Zhicheng Dou, Yutao Zhu, Ji-rong Wen

Large language models (LLMs) may generate text that lacks consistency with human knowledge, leading to factual inaccuracies or \textit{hallucination}. Existing research for evaluating the factuality of LLMs involves extracting fact claims using an LLM and verifying them against a predefined fact source. However, these evaluation metrics are task-specific, and not scalable, and the substitutability of fact sources in different tasks is under-explored. To address these challenges, we categorize four available fact sources: human-written evidence, reference documents, search engine results, and LLM knowledge, along with five text generation tasks containing six representative datasets. Then, we propose \texttt{UFO}, an LLM-based unified and flexible evaluation framework to verify facts against plug-and-play fact sources. We implement five evaluation scenarios based on this framework. Experimental results show that for most QA tasks, human-written evidence and reference documents are crucial, and they can substitute for each other in retrieval-augmented QA tasks. In news fact generation tasks, search engine results and LLM knowledge are essential. Our dataset and code are available at \url{https://github.com/WaldenRUC/UFO}.

http://arxiv.org/abs/2402.14690v1

cs.CL

Visual Hallucinations of Multi-modal Large Language Models

Wen Huang, Hongbin Liu, Minxin Guo, Neil Zhenqiang Gong

Visual hallucination (VH) means that a multi-modal LLM (MLLM) imagines incorrect details about an image in visual question answering. Existing studies find VH instances only in existing image datasets, which results in biased understanding of MLLMs' performance under VH due to limited diversity of such VH instances. In this work, we propose a tool called VHTest to generate a diverse set of VH instances. Specifically, VHTest finds some initial VH instances in existing image datasets (e.g., COCO), generates a text description for each VH mode, and uses a text-to-image generative model (e.g., DALL-E-3) to generate VH images based on the text descriptions. We collect a benchmark dataset with 1,200 VH instances in 8 VH modes using VHTest. We find that existing MLLMs such as GPT-4V, LLaVA-1.5, and MiniGPT-v2 hallucinate for a large fraction of the instances in our benchmark. Moreover, we find that fine-tuning an MLLM using our benchmark dataset reduces its likelihood to hallucinate without sacrificing its performance on other benchmarks. Our benchmarks are publicly available: https://github.com/wenhuang2000/VHTest.

http://arxiv.org/abs/2402.14683v1

cs.CV, cs.AI, cs.LG

ConceptMath: A Bilingual Concept-wise Benchmark for Measuring Mathematical Reasoning of Large Language Models

Yanan Wu, Jie Liu, Xingyuan Bu, Jiaheng Liu, Zhanhui Zhou, Yuanxing Zhang, Chenchen Zhang, Zhiqi Bai, Haibin Chen, Tiezheng Ge, Wanli Ouyang, Wenbo Su, Bo Zheng

This paper introduces ConceptMath, a bilingual (English and Chinese), fine-grained benchmark that evaluates concept-wise mathematical reasoning of Large Language Models (LLMs). Unlike traditional benchmarks that evaluate general mathematical reasoning with an average accuracy, ConceptMath systematically organizes math problems under a hierarchy of math concepts, so that mathematical reasoning can be evaluated at different granularity with concept-wise accuracies. Based on our ConcepthMath, we evaluate a broad range of LLMs, and we observe existing LLMs, though achieving high average accuracies on traditional benchmarks, exhibit significant performance variations across different math concepts and may even fail catastrophically on the most basic ones. Besides, we also introduce an efficient fine-tuning strategy to enhance the weaknesses of existing LLMs. Finally, we hope ConceptMath could guide the developers to understand the fine-grained mathematical abilities of their models and facilitate the growth of foundation models.

http://arxiv.org/abs/2402.14660v2

cs.CL, cs.AI

SIMPLOT: Enhancing Chart Question Answering by Distilling Essentials

Wonjoong Kim, Sangwu Park, Yeonjun In, Seokwon Han, Chanyoung Park

Recently, interpreting complex charts with logical reasoning have emerged as challenges due to the development of vision-language models. A prior state-of-the-art (SOTA) model, Deplot, has presented an end-to-end method that leverages the vision-language model to convert charts into table format utilizing Large Language Models (LLMs) for reasoning. However, unlike natural images, charts contain a mix of essential and irrelevant information required for chart reasoning, and we discover that this characteristic can lower the performance of chart-to-table extraction. In this paper, we introduce SIMPLOT, a method designed to extract only the elements necessary for chart reasoning. The proposed method involves two steps: 1) training to mimic a simple plot that contains only the essential information from a complex chart for table extraction, followed by 2) performing reasoning based on the table. Our model enables accurate chart reasoning without the need for additional annotations or datasets, and its effectiveness is demonstrated through various experiments. Furthermore, we propose a novel prompt addressing the shortcoming of recent SOTA model, ignoring visual attributes such as color. Our source code is available at https://github.com/sangwu99/Simplot.

http://arxiv.org/abs/2405.00021v1

cs.CV, cs.AI, cs.CL

Balanced Data Sampling for Language Model Training with Clustering

Yunfan Shao, Linyang Li, Zhaoye Fei, Hang Yan, Dahua Lin, Xipeng Qiu

Data plays a fundamental role in the training of Large Language Models (LLMs). While attention has been paid to the collection and composition of datasets, determining the data sampling strategy in training remains an open question. Most LLMs are trained with a simple strategy, random sampling. However, this sampling strategy ignores the unbalanced nature of training data distribution, which can be sub-optimal. In this paper, we propose ClusterClip Sampling to balance the text distribution of training data for better model training. Specifically, ClusterClip Sampling utilizes data clustering to reflect the data distribution of the training set and balances the common samples and rare samples during training based on the cluster results. A repetition clip operation is introduced to mitigate the overfitting issue led by samples from certain clusters. Extensive experiments validate the effectiveness of ClusterClip Sampling, which outperforms random sampling and other cluster-based sampling variants under various training datasets and large language models.

http://arxiv.org/abs/2402.14526v1

cs.CL, cs.AI

LLMBind: A Unified Modality-Task Integration Framework

Bin Zhu, Munan Ning, Peng Jin, Bin Lin, Jinfa Huang, Qi Song, Junwu Zhang, Zhenyu Tang, Mingjun Pan, Xing Zhou, Li Yuan

In the multi-modal domain, the dependence of various models on specific input formats leads to user confusion and hinders progress. To address this challenge, we introduce \textbf{LLMBind}, a novel framework designed to unify a diverse array of multi-modal tasks. By harnessing a Mixture-of-Experts (MoE) Large Language Model (LLM), LLMBind processes multi-modal inputs and generates task-specific tokens, enabling the invocation of corresponding models to accomplish tasks. This unique approach empowers LLMBind to interpret inputs and generate outputs across various modalities, including image, text, video, and audio. Furthermore, we have constructed an interaction dataset comprising 400k instructions, which unlocks the ability of LLMBind for interactive visual generation and editing tasks. Extensive experimentation demonstrates that LLMBind achieves very superior performance across diverse tasks and outperforms existing models in user evaluations conducted in real-world scenarios. Moreover, the adaptability of LLMBind allows for seamless integration with the latest models and extension to new modality tasks, highlighting its potential to serve as a unified AI agent for modeling universal modalities.

http://arxiv.org/abs/2402.14891v5

cs.CL, cs.AI

INSTRAUG: Automatic Instruction Augmentation for Multimodal Instruction Fine-tuning

Wei Han, Hui Chen, Soujanya Poria

Fine-tuning large language models (LLMs) on multi-task instruction-following data has been proven to be a powerful learning paradigm for improving their zero-shot capabilities on new tasks. Recent works about high-quality instruction-following data generation and selection require amounts of human labor to conceive model-understandable instructions for the given tasks and carefully filter the LLM-generated data. In this work, we introduce an automatic instruction augmentation method named INSTRAUG in multimodal tasks. It starts from a handful of basic and straightforward meta instructions but can expand an instruction-following dataset by 30 times. Results on two popular multimodal instructionfollowing benchmarks MULTIINSTRUCT and InstructBLIP show that INSTRAUG can significantly improve the alignment of multimodal large language models (MLLMs) across 12 multimodal tasks, which is even equivalent to the benefits of scaling up training data multiple times.

http://arxiv.org/abs/2402.14492v1

cs.CL, cs.AI

MeTMaP: Metamorphic Testing for Detecting False Vector Matching Problems in LLM Augmented Generation

Guanyu Wang, Yuekang Li, Yi Liu, Gelei Deng, Tianlin Li, Guosheng Xu, Yang Liu, Haoyu Wang, Kailong Wang

Augmented generation techniques such as Retrieval-Augmented Generation (RAG) and Cache-Augmented Generation (CAG) have revolutionized the field by enhancing large language model (LLM) outputs with external knowledge and cached information. However, the integration of vector databases, which serve as a backbone for these augmentations, introduces critical challenges, particularly in ensuring accurate vector matching. False vector matching in these databases can significantly compromise the integrity and reliability of LLM outputs, leading to misinformation or erroneous responses. Despite the crucial impact of these issues, there is a notable research gap in methods to effectively detect and address false vector matches in LLM-augmented generation. This paper presents MeTMaP, a metamorphic testing framework developed to identify false vector matching in LLM-augmented generation systems. We derive eight metamorphic relations (MRs) from six NLP datasets, which form our method's core, based on the idea that semantically similar texts should match and dissimilar ones should not. MeTMaP uses these MRs to create sentence triplets for testing, simulating real-world LLM scenarios. Our evaluation of MeTMaP over 203 vector matching configurations, involving 29 embedding models and 7 distance metrics, uncovers significant inaccuracies. The results, showing a maximum accuracy of only 41.51\% on our tests compared to the original datasets, emphasize the widespread issue of false matches in vector matching methods and the critical need for effective detection and mitigation in LLM-augmented applications.

http://arxiv.org/abs/2402.14480v1

cs.SE

Data Science with LLMs and Interpretable Models

Sebastian Bordt, Ben Lengerich, Harsha Nori, Rich Caruana

Recent years have seen important advances in the building of interpretable models, machine learning models that are designed to be easily understood by humans. In this work, we show that large language models (LLMs) are remarkably good at working with interpretable models, too. In particular, we show that LLMs can describe, interpret, and debug Generalized Additive Models (GAMs). Combining the flexibility of LLMs with the breadth of statistical patterns accurately described by GAMs enables dataset summarization, question answering, and model critique. LLMs can also improve the interaction between domain experts and interpretable models, and generate hypotheses about the underlying phenomenon. We release \url{https://github.com/interpretml/TalkToEBM} as an open-source LLM-GAM interface.

http://arxiv.org/abs/2402.14474v1

cs.LG, cs.CL

Do LLMs Implicitly Determine the Suitable Text Difficulty for Users?

Seiji Gobara, Hidetaka Kamigaito, Taro Watanabe

Education that suits the individual learning level is necessary to improve students' understanding. The first step in achieving this purpose by using large language models (LLMs) is to adjust the textual difficulty of the response to students. This work analyzes how LLMs can implicitly adjust text difficulty between user input and its generated text. To conduct the experiments, we created a new dataset from Stack-Overflow to explore the performance of question-answering-based conversation. Experimental results on the Stack-Overflow dataset and the TSCC dataset, including multi-turn conversation show that LLMs can implicitly handle text difficulty between user input and its generated response. We also observed that some LLMs can surpass humans in handling text difficulty and the importance of instruction-tuning.

http://arxiv.org/abs/2402.14453v1

cs.CL

COBIAS: Contextual Reliability in Bias Assessment

Priyanshul Govil, Vamshi Krishna Bonagiri, Manas Gaur, Ponnurangam Kumaraguru, Sanorita Dey

Large Language Models (LLMs) are trained on inherently biased data. Previous works on debiasing models rely on benchmark datasets to measure model performance. However, these datasets suffer from several pitfalls due to the extremely subjective understanding of bias, highlighting a critical need for contextual exploration. We propose understanding the context of user inputs with consideration of the diverse situations in which input statements are possible. This approach would allow for frameworks that foster bias awareness rather than guardrails that hurt user engagement. Our contribution is twofold: (i) we create a dataset of 2287 stereotyped statements augmented with points for adding context; (ii) we develop the Context-Oriented Bias Indicator and Assessment Score (COBIAS) to assess statements' contextual reliability in measuring bias. Our metric is a significant predictor of the contextual reliability of bias-benchmark datasets ($\chi^2=71.02, p<2.2 \cdot 10^{-16})$. COBIAS can be used to create reliable datasets, resulting in an improvement in bias mitigation works.

http://arxiv.org/abs/2402.14889v1

cs.CL, cs.AI

Enhancing Temporal Knowledge Graph Forecasting with Large Language Models via Chain-of-History Reasoning

Yuwei Xia, Ding Wang, Qiang Liu, Liang Wang, Shu Wu, Xiaoyu Zhang

Temporal Knowledge Graph (TKG) forecasting aims to predict future facts based on given histories. Most recent graph-based models excel at capturing structural information within TKGs but lack semantic comprehension abilities. Nowadays, with the surge of LLMs, the LLM-based TKG prediction model has emerged. However, the existing LLM-based model exhibits three shortcomings: (1) It only focuses on the first-order history for prediction while ignoring high-order historical information, resulting in the provided information for LLMs being extremely limited. (2) LLMs struggle with optimal reasoning performance under heavy historical information loads. (3) For TKG prediction, the temporal reasoning capability of LLM alone is limited. To address the first two challenges, we propose Chain-of-History (CoH) reasoning which explores high-order histories step-by-step, achieving effective utilization of high-order historical information for LLMs on TKG prediction. To address the third issue, we design CoH as a paly-and-plug module to enhance the performance of graph-based models for TKG prediction. Extensive experiments on three datasets and backbones demonstrate the effectiveness of CoH.

http://arxiv.org/abs/2402.14382v1

cs.CL

Small Language Model Is a Good Guide for Large Language Model in Chinese Entity Relation Extraction

Xuemei Tang, Jun Wang, Qi Su

Recently, large language models (LLMs) have been successful in relational extraction (RE) tasks, especially in the few-shot learning. An important problem in the field of RE is long-tailed data, while not much attention is currently paid to this problem using LLM approaches. Therefore, in this paper, we propose SLCoLM, a model collaboration framework, to mitigate the data long-tail problem. In our framework, We use the ``\textit{Training-Guide-Predict}'' strategy to combine the strengths of pre-trained language models (PLMs) and LLMs, where a task-specific PLM framework acts as a tutor, transfers task knowledge to the LLM, and guides the LLM in performing RE tasks. Our experiments on a RE dataset rich in relation types show that the approach in this paper facilitates RE of long-tail relation types.

http://arxiv.org/abs/2402.14373v1

cs.CL

Rethinking Scientific Summarization Evaluation: Grounding Explainable Metrics on Facet-aware Benchmark

Xiuying Chen, Tairan Wang, Qingqing Zhu, Taicheng Guo, Shen Gao, Zhiyong Lu, Xin Gao, Xiangliang Zhang

The summarization capabilities of pretrained and large language models (LLMs) have been widely validated in general areas, but their use in scientific corpus, which involves complex sentences and specialized knowledge, has been less assessed. This paper presents conceptual and experimental analyses of scientific summarization, highlighting the inadequacies of traditional evaluation methods, such as $n$-gram, embedding comparison, and QA, particularly in providing explanations, grasping scientific concepts, or identifying key content. Subsequently, we introduce the Facet-aware Metric (FM), employing LLMs for advanced semantic matching to evaluate summaries based on different aspects. This facet-aware approach offers a thorough evaluation of abstracts by decomposing the evaluation task into simpler subtasks.Recognizing the absence of an evaluation benchmark in this domain, we curate a Facet-based scientific summarization Dataset (FD) with facet-level annotations. Our findings confirm that FM offers a more logical approach to evaluating scientific summaries. In addition, fine-tuned smaller models can compete with LLMs in scientific contexts, while LLMs have limitations in learning from in-context information in scientific domains. This suggests an area for future enhancement of LLMs.

http://arxiv.org/abs/2402.14359v1

cs.CL

Rule or Story, Which is a Better Commonsense Expression for Talking with Large Language Models?

Ning Bian, Xianpei Han, Hongyu Lin, Yaojie Lu, Ben He, Le Sun

Building machines with commonsense has been a longstanding challenge in NLP due to the reporting bias of commonsense rules and the exposure bias of rule-based commonsense reasoning. In contrast, humans convey and pass down commonsense implicitly through stories. This paper investigates the inherent commonsense ability of large language models (LLMs) expressed through storytelling. We systematically investigate and compare stories and rules for retrieving and leveraging commonsense in LLMs. Experimental results on 28 commonsense QA datasets show that stories outperform rules as the expression for retrieving commonsense from LLMs, exhibiting higher generation confidence and commonsense accuracy. Moreover, stories are the more effective commonsense expression for answering questions regarding daily events, while rules are more effective for scientific questions. This aligns with the reporting bias of commonsense in text corpora. We further show that the correctness and relevance of commonsense stories can be further improved via iterative self-supervised fine-tuning. These findings emphasize the importance of using appropriate language to express, retrieve, and leverage commonsense for LLMs, highlighting a promising direction for better exploiting their commonsense abilities.

http://arxiv.org/abs/2402.14355v1

cs.CL

Triad: A Framework Leveraging a Multi-Role LLM-based Agent to Solve Knowledge Base Question Answering

Chang Zong, Yuchen Yan, Weiming Lu, Jian Shao, Eliot Huang, Heng Chang, Yueting Zhuang

Recent progress with LLM-based agents has shown promising results across various tasks. However, their use in answering questions from knowledge bases remains largely unexplored. Implementing a KBQA system using traditional methods is challenging due to the shortage of task-specific training data and the complexity of creating task-focused model structures. In this paper, we present Triad, a unified framework that utilizes an LLM-based agent with three roles for KBQA tasks. The agent is assigned three roles to tackle different KBQA subtasks: agent as a generalist for mastering various subtasks, as a decision maker for the selection of candidates, and as an advisor for answering questions with knowledge. Our KBQA framework is executed in four phases, involving the collaboration of the agent's multiple roles. We evaluated the performance of our framework using three benchmark datasets, and the results show that our framework outperforms state-of-the-art systems on the LC-QuAD and YAGO-QA benchmarks, yielding F1 scores of 11.8% and 20.7%, respectively.

http://arxiv.org/abs/2402.14320v5

cs.CL, cs.AI, 68T50, I.2.7

Hint-before-Solving Prompting: Guiding LLMs to Effectively Utilize Encoded Knowledge

Jinlan Fu, Shenzhen Huangfu, Hang Yan, See-Kiong Ng, Xipeng Qiu

Large Language Models (LLMs) have recently showcased remarkable generalizability in various domains. Despite their extensive knowledge, LLMs still face challenges in efficiently utilizing encoded knowledge to develop accurate and logical reasoning processes. To mitigate this problem, we introduced Hint-before-Solving Prompting (HSP), which guides the model to generate hints (e.g., specific knowledge or key ideas) for solving the problem and then generate solutions containing intermediate reasoning steps. Since HSP is orthogonal to prompting methods (e.g., Chain-of-Thought (CoT)), we applied HSP to CoT, Least-to-Most, Plan-and-Solve, and Standard promptings. The results of extensive experiments on 6 reasoning benchmarks and 4 open-source LLMs demonstrate that HSP can effectively improve the accuracy of reasoning tasks: (1) By applying high-quality hint-enhanced HSP to CoT prompting, Llama2-70B-Chat shows an improvement of 9.7. (2) Beyond exploring training-free LLM capabilities, we built the HSPMATH dataset based on HSP and fine-tuned Llemma-7B, reaching 64.3 accuracy, surpassing GPT-3.5 and WizardMath-13B. We make our code and dataset publicly available at \url{https://github.com/jinlanfu/HSP}.

http://arxiv.org/abs/2402.14310v1

cs.CL

Double-I Watermark: Protecting Model Copyright for LLM Fine-tuning

Shen Li, Liuyi Yao, Jinyang Gao, Lan Zhang, Yaliang Li

To support various applications, business owners often seek the customized models that are obtained by fine-tuning a pre-trained LLM through the API provided by LLM owners or cloud servers. However, this process carries a substantial risk of model misuse, potentially resulting in severe economic consequences for business owners. Thus, safeguarding the copyright of these customized models during LLM fine-tuning has become an urgent practical requirement, but there are limited existing solutions to provide such protection. To tackle this pressing issue, we propose a novel watermarking approach named "Double-I watermark". Specifically, based on the instruct-tuning data, two types of backdoor data paradigms are introduced with trigger in the instruction and the input, respectively. By leveraging LLM's learning capability to incorporate customized backdoor samples into the dataset, the proposed approach effectively injects specific watermarking information into the customized model during fine-tuning, which makes it easy to inject and verify watermarks in commercial scenarios. We evaluate the proposed "Double-I watermark" under various fine-tuning methods, demonstrating its harmlessness, robustness, uniqueness, imperceptibility, and validity through both theoretical analysis and experimental verification.

http://arxiv.org/abs/2402.14883v1

cs.CR, cs.AI, cs.LG

Take the Bull by the Horns: Hard Sample-Reweighted Continual Training Improves LLM Generalization

Xuxi Chen, Zhendong Wang, Daouda Sow, Junjie Yang, Tianlong Chen, Yingbin Liang, Mingyuan Zhou, Zhangyang Wang

In the rapidly advancing arena of large language models (LLMs), a key challenge is to enhance their capabilities amid a looming shortage of high-quality training data. Our study starts from an empirical strategy for the light continual training of LLMs using their original pre-training data sets, with a specific focus on selective retention of samples that incur moderately high losses. These samples are deemed informative and beneficial for model refinement, contrasting with the highest-loss samples, which would be discarded due to their correlation with data noise and complexity. We then formalize this strategy into a principled framework of Instance-Reweighted Distributionally Robust Optimization (IR-DRO). IR-DRO is designed to dynamically prioritize the training focus on informative samples through an instance reweighting mechanism, streamlined by a closed-form solution for straightforward integration into established training protocols. Through rigorous experimentation with various models and datasets, our findings indicate that our sample-targeted methods significantly improve LLM performance across multiple benchmarks, in both continual pre-training and instruction tuning scenarios. Our codes are available at https://github.com/VITA-Group/HardFocusTraining.

http://arxiv.org/abs/2402.14270v2

cs.LG

Can Large Language Models Detect Misinformation in Scientific News Reporting?

Yupeng Cao, Aishwarya Muralidharan Nair, Elyon Eyimife, Nastaran Jamalipour Soofi, K. P. Subbalakshmi, John R. Wullert II, Chumki Basu, David Shallcross

Scientific facts are often spun in the popular press with the intent to influence public opinion and action, as was evidenced during the COVID-19 pandemic. Automatic detection of misinformation in the scientific domain is challenging because of the distinct styles of writing in these two media types and is still in its nascence. Most research on the validity of scientific reporting treats this problem as a claim verification challenge. In doing so, significant expert human effort is required to generate appropriate claims. Our solution bypasses this step and addresses a more real-world scenario where such explicit, labeled claims may not be available. The central research question of this paper is whether it is possible to use large language models (LLMs) to detect misinformation in scientific reporting. To this end, we first present a new labeled dataset SciNews, containing 2.4k scientific news stories drawn from trusted and untrustworthy sources, paired with related abstracts from the CORD-19 database. Our dataset includes both human-written and LLM-generated news articles, making it more comprehensive in terms of capturing the growing trend of using LLMs to generate popular press articles. Then, we identify dimensions of scientific validity in science news articles and explore how this can be integrated into the automated detection of scientific misinformation. We propose several baseline architectures using LLMs to automatically detect false representations of scientific findings in the popular press. For each of these architectures, we use several prompt engineering strategies including zero-shot, few-shot, and chain-of-thought prompting. We also test these architectures and prompting strategies on GPT-3.5, GPT-4, and Llama2-7B, Llama2-13B.

http://arxiv.org/abs/2402.14268v1

cs.CL, cs.AI, cs.SI

Word-Sequence Entropy: Towards Uncertainty Estimation in Free-Form Medical Question Answering Applications and Beyond

Zhiyuan Wang, Jinhao Duan, Chenxi Yuan, Qingyu Chen, Tianlong Chen, Huaxiu Yao, Yue Zhang, Ren Wang, Kaidi Xu, Xiaoshuang Shi

Uncertainty estimation plays a pivotal role in ensuring the reliability of safety-critical human-AI interaction systems, particularly in the medical domain. However, a general method for quantifying the uncertainty of free-form answers has yet to be established in open-ended medical question-answering (QA) tasks, where irrelevant words and sequences with limited semantic information can be the primary source of uncertainty due to the presence of generative inequality. In this paper, we propose the Word-Sequence Entropy (WSE), which calibrates the uncertainty proportion at both the word and sequence levels according to the semantic relevance, with greater emphasis placed on keywords and more relevant sequences when performing uncertainty quantification. We compare WSE with 6 baseline methods on 5 free-form medical QA datasets, utilizing 7 "off-the-shelf" large language models (LLMs), and show that WSE exhibits superior performance on accurate uncertainty measurement under two standard criteria for correctness evaluation (e.g., WSE outperforms existing state-of-the-art method by 3.23% AUROC on the MedQA dataset). Additionally, in terms of the potential for real-world medical QA applications, we achieve a significant enhancement in the performance of LLMs when employing sequences with lower uncertainty, identified by WSE, as final answers (e.g., +6.36% accuracy improvement on the COVID-QA dataset), without requiring any additional task-specific fine-tuning or architectural modifications.

http://arxiv.org/abs/2402.14259v1

cs.CL, cs.AI, cs.LG

Eagle: Ethical Dataset Given from Real Interactions

Masahiro Kaneko, Danushka Bollegala, Timothy Baldwin

Recent studies have demonstrated that large language models (LLMs) have ethical-related problems such as social biases, lack of moral reasoning, and generation of offensive content. The existing evaluation metrics and methods to address these ethical challenges use datasets intentionally created by instructing humans to create instances including ethical problems. Therefore, the data does not reflect prompts that users actually provide when utilizing LLM services in everyday contexts. This may not lead to the development of safe LLMs that can address ethical challenges arising in real-world applications. In this paper, we create Eagle datasets extracted from real interactions between ChatGPT and users that exhibit social biases, toxicity, and immoral problems. Our experiments show that Eagle captures complementary aspects, not covered by existing datasets proposed for evaluation and mitigation of such ethical challenges. Our code is publicly available at https://huggingface.co/datasets/MasahiroKaneko/eagle.

http://arxiv.org/abs/2402.14258v1

cs.CL

Learning to Reduce: Optimal Representations of Structured Data in Prompting Large Language Models

Younghun Lee, Sungchul Kim, Tong Yu, Ryan A. Rossi, Xiang Chen

Large Language Models (LLMs) have been widely used as general-purpose AI agents showing comparable performance on many downstream tasks. However, existing work shows that it is challenging for LLMs to integrate structured data (e.g. KG, tables, DBs) into their prompts; LLMs need to either understand long text data or select the most relevant evidence prior to inference, and both approaches are not trivial. In this paper, we propose a framework, Learning to Reduce, that fine-tunes a language model to generate a reduced version of an input context, given a task description and context input. The model learns to reduce the input context using On-Policy Reinforcement Learning and aims to improve the reasoning performance of a fixed LLM. Experimental results illustrate that our model not only achieves comparable accuracies in selecting the relevant evidence from an input context, but also shows generalizability on different datasets. We further show that our model helps improve the LLM's performance on downstream tasks especially when the context is long.

http://arxiv.org/abs/2402.14195v1

cs.CL

Understanding the Dataset Practitioners Behind Large Language Model Development

Crystal Qian, Emily Reif, Minsuk Kahng

As large language models (LLMs) become more advanced and impactful, it is increasingly important to scrutinize the data that they rely upon and produce. What is it to be a dataset practitioner doing this work? We approach this in two parts: first, we define the role of "dataset practitioners" by performing a retrospective analysis on the responsibilities of teams contributing to LLM development at a technology company, Google. Then, we conduct semi-structured interviews with a cross-section of these practitioners (N=10). We find that although data quality is a top priority, there is little consensus around what data quality is and how to evaluate it. Consequently, practitioners either rely on their own intuition or write custom code to evaluate their data. We discuss potential reasons for this phenomenon and opportunities for alignment.

http://arxiv.org/abs/2402.16611v2

cs.CL, cs.AI, cs.HC

Automatic Histograms: Leveraging Language Models for Text Dataset Exploration

Emily Reif, Crystal Qian, James Wexler, Minsuk Kahng

Making sense of unstructured text datasets is perennially difficult, yet increasingly relevant with Large Language Models. Data workers often rely on dataset summaries, especially distributions of various derived features. Some features, like toxicity or topics, are relevant to many datasets, but many interesting features are domain specific: instruments and genres for a music dataset, or diseases and symptoms for a medical dataset. Accordingly, data workers often run custom analyses for each dataset, which is cumbersome and difficult. We present AutoHistograms, a visualization tool leveragingLLMs. AutoHistograms automatically identifies relevant features, visualizes them with histograms, and allows the user to interactively query the dataset for categories of entities and create new histograms. In a user study with 10 data workers (n=10), we observe that participants can quickly identify insights and explore the data using AutoHistograms, and conceptualize a broad range of applicable use cases. Together, this tool and user study contributeto the growing field of LLM-assisted sensemaking tools.

http://arxiv.org/abs/2402.14880v1

cs.CL, cs.AI, cs.HC

DeiSAM: Segment Anything with Deictic Prompting

Hikaru Shindo, Manuel Brack, Gopika Sudhakaran, Devendra Singh Dhami, Patrick Schramowski, Kristian Kersting

Large-scale, pre-trained neural networks have demonstrated strong capabilities in various tasks, including zero-shot image segmentation. To identify concrete objects in complex scenes, humans instinctively rely on deictic descriptions in natural language, i.e., referring to something depending on the context such as "The object that is on the desk and behind the cup.". However, deep learning approaches cannot reliably interpret such deictic representations due to their lack of reasoning capabilities in complex scenarios. To remedy this issue, we propose DeiSAM -- a combination of large pre-trained neural networks with differentiable logic reasoners -- for deictic promptable segmentation. Given a complex, textual segmentation description, DeiSAM leverages Large Language Models (LLMs) to generate first-order logic rules and performs differentiable forward reasoning on generated scene graphs. Subsequently, DeiSAM segments objects by matching them to the logically inferred image regions. As part of our evaluation, we propose the Deictic Visual Genome (DeiVG) dataset, containing paired visual input and complex, deictic textual prompts. Our empirical results demonstrate that DeiSAM is a substantial improvement over purely data-driven baselines for deictic promptable segmentation.

http://arxiv.org/abs/2402.14123v1

cs.LG, cs.AI, cs.CV

FanOutQA: Multi-Hop, Multi-Document Question Answering for Large Language Models

Andrew Zhu, Alyssa Hwang, Liam Dugan, Chris Callison-Burch

One type of question that is commonly found in day-to-day scenarios is ``fan-out'' questions, complex multi-hop, multi-document reasoning questions that require finding information about a large number of entities. However, there exist few resources to evaluate this type of question-answering capability among large language models. To evaluate complex reasoning in LLMs more fully, we present FanOutQA, a high-quality dataset of fan-out question-answer pairs and human-annotated decompositions with English Wikipedia as the knowledge base. We formulate three benchmark settings across our dataset and benchmark 7 LLMs, including GPT-4, LLaMA 2, Claude-2.1, and Mixtral-8x7B, finding that contemporary models still have room to improve reasoning over inter-document dependencies in a long context. We provide our dataset and open-source tools to run models to encourage evaluation at https://fanoutqa.com

http://arxiv.org/abs/2402.14116v1

cs.CL, cs.AI

Diet-ODIN: A Novel Framework for Opioid Misuse Detection with Interpretable Dietary Patterns

Zheyuan Zhang, Zehong Wang, Shifu Hou, Evan Hall, Landon Bachman, Vincent Galassi, Jasmine White, Nitesh V. Chawla, Chuxu Zhang, Yanfang Ye

The opioid crisis has been one of the most critical society concerns in the United States. Although the medication assisted treatment (MAT) is recognized as the most effective treatment for opioid misuse and addiction, the various side effects can trigger opioid relapse. In addition to MAT, the dietary nutrition intervention has been demonstrated its importance in opioid misuse prevention and recovery. However, research on the alarming connections between dietary patterns and opioid misuse remain under-explored. In response to this gap, in this paper, we first establish a large-scale multifaceted dietary benchmark dataset related to opioid users at the first attempt and then develop a novel framework - i.e., namely Opioid Misuse Detection with Interpretable Dietary Patterns (Diet-ODIN) - to bridge heterogeneous graph (HG) and large language model (LLM) for the identification of users with opioid misuse and the interpretation of their associated dietary patterns. Specifically, in Diet-ODIN, we first construct an HG to comprehensively incorporate both dietary and health-related information, and then we devise a holistic graph learning framework with noise reduction to fully capitalize both users' individual dietary habits and shared dietary patterns for the detection of users with opioid misuse. To further delve into the intricate correlations between dietary patterns and opioid misuse, we exploit an LLM by utilizing the knowledge obtained from the graph learning model for interpretation. The extensive experimental results based on our established benchmark with quantitative and qualitative measures demonstrate the outstanding performance of Diet-ODIN in exploring the complex interplay between opioid misuse and dietary patterns, by comparison with state-of-the-art baseline methods.

http://arxiv.org/abs/2403.08820v1

cs.LG, cs.AI, cs.SI

Towards Building Multilingual Language Model for Medicine

Pengcheng Qiu, Chaoyi Wu, Xiaoman Zhang, Weixiong Lin, Haicheng Wang, Ya Zhang, Yanfeng Wang, Weidi Xie

In this paper, we aim to develop an open-source, multilingual language model for medicine, that the benefits a wider, linguistically diverse audience from different regions. In general, we present the contribution from the following aspects: first, for multilingual medical-specific adaptation, we construct a new multilingual medical corpus, that contains approximately 25.5B tokens encompassing 6 main languages, termed as MMedC, that enables auto-regressive training for existing general LLMs. second, to monitor the development of multilingual LLMs in medicine, we propose a new multilingual medical multi-choice question-answering benchmark with rationale, termed as MMedBench; third, we have assessed a number of popular, opensource large language models (LLMs) on our benchmark, along with those further auto-regressive trained on MMedC, as a result, our final model, termed as MMedLM 2, with only 7B parameters, achieves superior performance compared to all other open-source models, even rivaling GPT-4 on MMedBench. We will make the resources publicly available, including code, model weights, and datasets.

http://arxiv.org/abs/2402.13963v2

cs.CL

Distillation Contrastive Decoding: Improving LLMs Reasoning with Contrastive Decoding and Distillation

Phuc Phan, Hieu Tran, Long Phan

We propose a straightforward approach called Distillation Contrastive Decoding (DCD) to enhance the reasoning capabilities of Large Language Models (LLMs) during inference. In contrast to previous approaches that relied on smaller amateur models or analysis of hidden state differences, DCD employs Contrastive Chain-of-thought Prompting and advanced distillation techniques, including Dropout and Quantization. This approach effectively addresses the limitations of Contrastive Decoding (CD), which typically requires both an expert and an amateur model, thus increasing computational resource demands. By integrating contrastive prompts with distillation, DCD obviates the need for an amateur model and reduces memory usage. Our evaluations demonstrate that DCD significantly enhances LLM performance across a range of reasoning benchmarks, surpassing both CD and existing methods in the GSM8K and StrategyQA datasets.

http://arxiv.org/abs/2402.14874v1

cs.CL, cs.AI, cs.LG

Calibrating Large Language Models with Sample Consistency

Qing Lyu, Kumar Shridhar, Chaitanya Malaviya, Li Zhang, Yanai Elazar, Niket Tandon, Marianna Apidianaki, Mrinmaya Sachan, Chris Callison-Burch

Accurately gauging the confidence level of Large Language Models' (LLMs) predictions is pivotal for their reliable application. However, LLMs are often uncalibrated inherently and elude conventional calibration techniques due to their proprietary nature and massive scale. In this work, we explore the potential of deriving confidence from the distribution of multiple randomly sampled model generations, via three measures of consistency. We perform an extensive evaluation across various open and closed-source models on nine reasoning datasets. Results show that consistency-based calibration methods outperform existing post-hoc approaches. Meanwhile, we find that factors such as intermediate explanations, model scaling, and larger sample sizes enhance calibration, while instruction-tuning makes calibration more difficult. Moreover, confidence scores obtained from consistency have the potential to enhance model performance. Finally, we offer practical guidance on choosing suitable consistency metrics for calibration, tailored to the characteristics of various LMs.

http://arxiv.org/abs/2402.13904v1

cs.CL

An Explainable Transformer-based Model for Phishing Email Detection: A Large Language Model Approach

Mohammad Amaz Uddin, Iqbal H. Sarker

Phishing email is a serious cyber threat that tries to deceive users by sending false emails with the intention of stealing confidential information or causing financial harm. Attackers, often posing as trustworthy entities, exploit technological advancements and sophistication to make detection and prevention of phishing more challenging. Despite extensive academic research, phishing detection remains an ongoing and formidable challenge in the cybersecurity landscape. Large Language Models (LLMs) and Masked Language Models (MLMs) possess immense potential to offer innovative solutions to address long-standing challenges. In this research paper, we present an optimized, fine-tuned transformer-based DistilBERT model designed for the detection of phishing emails. In the detection process, we work with a phishing email dataset and utilize the preprocessing techniques to clean and solve the imbalance class issues. Through our experiments, we found that our model effectively achieves high accuracy, demonstrating its capability to perform well. Finally, we demonstrate our fine-tuned model using Explainable-AI (XAI) techniques such as Local Interpretable Model-Agnostic Explanations (LIME) and Transformer Interpret to explain how our model makes predictions in the context of text classification for phishing emails.

http://arxiv.org/abs/2402.13871v1

cs.LG, cs.AI, cs.CR

Kuaiji: the First Chinese Accounting Large Language Model

Jiayuan Luo, Songhua Yang, Xiaoling Qiu, Panyu Chen, Yufei Nai, Wenxuan Zeng, Wentao Zhang, Xinke Jiang

Large Language Models (LLMs) like ChatGPT and GPT-4 have demonstrated impressive proficiency in comprehending and generating natural language. However, they encounter difficulties when tasked with adapting to specialized domains such as accounting. To address this challenge, we introduce Kuaiji, a tailored Accounting Large Language Model. Kuaiji is meticulously fine-tuned using the Baichuan framework, which encompasses continuous pre-training and supervised fine-tuning processes. Supported by CAtAcctQA, a dataset containing large genuine accountant-client dialogues, Kuaiji exhibits exceptional accuracy and response speed. Our contributions encompass the creation of the first Chinese accounting dataset, the establishment of Kuaiji as a leading open-source Chinese accounting LLM, and the validation of its efficacy through real-world accounting scenarios.

http://arxiv.org/abs/2402.13866v2

cs.CL, cs.AI

LLM4SBR: A Lightweight and Effective Framework for Integrating Large Language Models in Session-based Recommendation

Shutong Qiao, Chen Gao, Junhao Wen, Wei Zhou, Qun Luo, Peixuan Chen, Yong Li

Traditional session-based recommendation (SBR) utilizes session behavior sequences from anonymous users for recommendation. Although this strategy is highly efficient, it sacrifices the inherent semantic information of the items, making it difficult for the model to understand the true intent of the session and resulting in a lack of interpretability in the recommended results. Recently, large language models (LLMs) have flourished across various domains, offering a glimpse of hope in addressing the aforementioned challenges. Inspired by the impact of LLMs, research exploring the integration of LLMs with the Recommender system (RS) has surged like mushrooms after rain. However, constrained by high time and space costs, as well as the brief and anonymous nature of session data, the first LLM recommendation framework suitable for industrial deployment has yet to emerge in the field of SBR. To address the aforementioned challenges, we have proposed the LLM Integration Framework for SBR (LLM4SBR). Serving as a lightweight and plug-and-play framework, LLM4SBR adopts a two-step strategy. Firstly, we transform session data into a bimodal form of text and behavior. In the first step, leveraging the inferential capabilities of LLMs, we conduct inference on session text data from different perspectives and design the component for auxiliary enhancement. In the second step, the SBR model is trained on behavior data, aligning and averaging two modal session representations from different perspectives. Finally, we fuse session representations from different perspectives and modalities as the ultimate session representation for recommendation. We conducted experiments on two real-world datasets, and the results demonstrate that LLM4SBR significantly improves the performance of traditional SBR models and is highly lightweight and efficient, making it suitable for industrial deployment.

http://arxiv.org/abs/2402.13840v1

cs.IR, cs.AI

CriticBench: Evaluating Large Language Models as Critic

Tian Lan, Wenwei Zhang, Chen Xu, Heyan Huang, Dahua Lin, Kai Chen, Xian-ling Mao

Critique ability are crucial in the scalable oversight and self-improvement of Large Language Models (LLMs). While many recent studies explore the critique ability of LLMs to judge and refine flaws in generations, how to comprehensively and reliably measure the critique abilities of LLMs is under-explored. This paper introduces CriticBench, a novel benchmark designed to comprehensively and reliably evaluate four key critique ability dimensions of LLMs: feedback, comparison, refinement and meta-feedback. CriticBench encompasses nine diverse tasks, each assessing the LLMs' ability to critique responses at varying levels of quality granularity. Our extensive evaluations of open-source and closed-source LLMs reveal intriguing relationships between the critique ability and tasks, response qualities, and model scales. Datasets, resources and evaluation toolkit for CriticBench will be publicly released at https://github.com/open-compass/CriticBench.

http://arxiv.org/abs/2402.13764v3

cs.CL, cs.AI

Factual Consistency Evaluation of Summarisation in the Era of Large Language Models

Zheheng Luo, Qianqian Xie, Sophia Ananiadou

Factual inconsistency with source documents in automatically generated summaries can lead to misinformation or pose risks. Existing factual consistency(FC) metrics are constrained by their performance, efficiency, and explainability. Recent advances in Large language models (LLMs) have demonstrated remarkable potential in text evaluation but their effectiveness in assessing FC in summarisation remains underexplored. Prior research has mostly focused on proprietary LLMs, leaving essential factors that affect their assessment capabilities unexplored. Additionally, current FC evaluation benchmarks are restricted to news articles, casting doubt on the generality of the FC methods tested on them. In this paper, we first address the gap by introducing TreatFact a dataset of LLM-generated summaries of clinical texts, annotated for FC by domain experts. Moreover, we benchmark 11 LLMs for FC evaluation across news and clinical domains and analyse the impact of model size, prompts, pre-training and fine-tuning data. Our findings reveal that despite proprietary models prevailing on the task, open-source LLMs lag behind. Nevertheless, there is potential for enhancing the performance of open-source LLMs through increasing model size, expanding pre-training data, and developing well-curated fine-tuning data. Experiments on TreatFact suggest that both previous methods and LLM-based evaluators are unable to capture factual inconsistencies in clinical summaries, posing a new challenge for FC evaluation.

http://arxiv.org/abs/2402.13758v1

cs.CL

Breaking the Barrier: Utilizing Large Language Models for Industrial Recommendation Systems through an Inferential Knowledge Graph

Qian Zhao, Hao Qian, Ziqi Liu, Gong-Duo Zhang, Lihong Gu

Recommendation systems are widely used in e-commerce websites and online platforms to address information overload. However, existing systems primarily rely on historical data and user feedback, making it difficult to capture user intent transitions. Recently, Knowledge Base (KB)-based models are proposed to incorporate expert knowledge, but it struggle to adapt to new items and the evolving e-commerce environment. To address these challenges, we propose a novel Large Language Model based Complementary Knowledge Enhanced Recommendation System (LLM-KERec). It introduces an entity extractor that extracts unified concept terms from item and user information. To provide cost-effective and reliable prior knowledge, entity pairs are generated based on entity popularity and specific strategies. The large language model determines complementary relationships in each entity pair, constructing a complementary knowledge graph. Furthermore, a new complementary recall module and an Entity-Entity-Item (E-E-I) weight decision model refine the scoring of the ranking model using real complementary exposure-click samples. Extensive experiments conducted on three industry datasets demonstrate the significant performance improvement of our model compared to existing approaches. Additionally, detailed analysis shows that LLM-KERec enhances users' enthusiasm for consumption by recommending complementary items. In summary, LLM-KERec addresses the limitations of traditional recommendation systems by incorporating complementary knowledge and utilizing a large language model to capture user intent transitions, adapt to new items, and enhance recommendation efficiency in the evolving e-commerce landscape.

http://arxiv.org/abs/2402.13750v1

cs.IR, cs.AI, cs.CL

From Text to CQL: Bridging Natural Language and Corpus Search Engine

Luming Lu, Jiyuan An, Yujie Wang, Liner yang, Cunliang Kong, Zhenghao Liu, Shuo Wang, Haozhe Lin, Mingwei Fang, Yaping Huang, Erhong Yang

Natural Language Processing (NLP) technologies have revolutionized the way we interact with information systems, with a significant focus on converting natural language queries into formal query languages such as SQL. However, less emphasis has been placed on the Corpus Query Language (CQL), a critical tool for linguistic research and detailed analysis within text corpora. The manual construction of CQL queries is a complex and time-intensive task that requires a great deal of expertise, which presents a notable challenge for both researchers and practitioners. This paper presents the first text-to-CQL task that aims to automate the translation of natural language into CQL. We present a comprehensive framework for this task, including a specifically curated large-scale dataset and methodologies leveraging large language models (LLMs) for effective text-to-CQL task. In addition, we established advanced evaluation metrics to assess the syntactic and semantic accuracy of the generated queries. We created innovative LLM-based conversion approaches and detailed experiments. The results demonstrate the efficacy of our methods and provide insights into the complexities of text-to-CQL task.

http://arxiv.org/abs/2402.13740v1

cs.CL

SaGE: Evaluating Moral Consistency in Large Language Models

Vamshi Krishna Bonagiri, Sreeram Vennam, Priyanshul Govil, Ponnurangam Kumaraguru, Manas Gaur

Despite recent advancements showcasing the impressive capabilities of Large Language Models (LLMs) in conversational systems, we show that even state-of-the-art LLMs are morally inconsistent in their generations, questioning their reliability (and trustworthiness in general). Prior works in LLM evaluation focus on developing ground-truth data to measure accuracy on specific tasks. However, for moral scenarios that often lack universally agreed-upon answers, consistency in model responses becomes crucial for their reliability. To address this issue, we propose an information-theoretic measure called Semantic Graph Entropy (SaGE), grounded in the concept of "Rules of Thumb" (RoTs) to measure a model's moral consistency. RoTs are abstract principles learned by a model and can help explain their decision-making strategies effectively. To this extent, we construct the Moral Consistency Corpus (MCC), containing 50K moral questions, responses to them by LLMs, and the RoTs that these models followed. Furthermore, to illustrate the generalizability of SaGE, we use it to investigate LLM consistency on two popular datasets -- TruthfulQA and HellaSwag. Our results reveal that task-accuracy and consistency are independent problems, and there is a dire need to investigate these issues further.

http://arxiv.org/abs/2402.13709v2

cs.CL, cs.AI

Investigating Multilingual Instruction-Tuning: Do Polyglot Models Demand for Multilingual Instructions?

Alexander Arno Weber, Klaudia Thellmann, Jan Ebert, Nicolas Flores-Herr, Jens Lehmann, Michael Fromm, Mehdi Ali

The adaption of multilingual pre-trained Large Language Models (LLMs) into eloquent and helpful assistants is essential to facilitate their use across different language regions. In that spirit, we are the first to conduct an extensive study of the performance of multilingual models on parallel, multi-turn instruction-tuning benchmarks across a selection of the most-spoken Indo-European languages. We systematically examine the effects of language and instruction dataset size on a mid-sized, multilingual LLM by instruction-tuning it on parallel instruction-tuning datasets. Our results demonstrate that instruction-tuning on parallel instead of monolingual corpora benefits cross-lingual instruction following capabilities by up to 4.6%. Furthermore, we show that the Superficial Alignment Hypothesis does not hold in general, as the investigated multilingual 7B parameter model presents a counter-example requiring large-scale instruction-tuning datasets. Finally, we conduct a human annotation study to understand the alignment between human-based and GPT-4-based evaluation within multilingual chat scenarios.

http://arxiv.org/abs/2402.13703v1

cs.CL

Self-Distillation Bridges Distribution Gap in Language Model Fine-Tuning

Zhaorui Yang, Qian Liu, Tianyu Pang, Han Wang, Haozhe Feng, Minfeng Zhu, Wei Chen

The surge in Large Language Models (LLMs) has revolutionized natural language processing, but fine-tuning them for specific tasks often encounters challenges in balancing performance and preserving general instruction-following abilities. In this paper, we posit that the distribution gap between task datasets and the LLMs serves as the primary underlying cause. To address the problem, we introduce Self-Distillation Fine-Tuning (SDFT), a novel approach that bridges the distribution gap by guiding fine-tuning with a distilled dataset generated by the model itself to match its original distribution. Experimental results on the Llama-2-chat model across various benchmarks demonstrate that SDFT effectively mitigates catastrophic forgetting while achieving comparable or superior performance on downstream tasks compared to the vanilla fine-tuning. Moreover, SDFT demonstrates the potential to maintain the helpfulness and safety alignment of LLMs. Our code is available at \url{https://github.com/sail-sg/sdft}.

http://arxiv.org/abs/2402.13669v1

cs.CL

PQA: Zero-shot Protein Question Answering for Free-form Scientific Enquiry with Large Language Models

Eli M Carrami, Sahand Sharifzadeh

We introduce the novel task of zero-shot Protein Question Answering (PQA) for free-form scientific enquiry. Given a previously unseen protein sequence and a natural language question, the task is to deliver a scientifically accurate answer. This task not only supports future biological research, but could also provide a test bed for assessing the scientific precision of large language models (LLMs). We contribute the first specialized dataset for PQA model training, containing 257K protein sequences annotated with 1.97M scientific question-answer pairs. Additionally, we propose and study several novel biologically relevant benchmarks for scientific PQA. Employing two robust multi-modal architectures, we establish an initial state-of-the-art performance for PQA and reveal key performance factors through ablation studies. Our comprehensive PQA framework, named Pika, including dataset, code, model checkpoints, and a user-friendly demo, is openly accessible on github.com/EMCarrami/Pika, promoting wider research and application in the field.

http://arxiv.org/abs/2402.13653v1

cs.LG

Unsupervised Text Style Transfer via LLMs and Attention Masking with Multi-way Interactions

Lei Pan, Yunshi Lan, Yang Li, Weining Qian

Unsupervised Text Style Transfer (UTST) has emerged as a critical task within the domain of Natural Language Processing (NLP), aiming to transfer one stylistic aspect of a sentence into another style without changing its semantics, syntax, or other attributes. This task is especially challenging given the intrinsic lack of parallel text pairings. Among existing methods for UTST tasks, attention masking approach and Large Language Models (LLMs) are deemed as two pioneering methods. However, they have shortcomings in generating unsmooth sentences and changing the original contents, respectively. In this paper, we investigate if we can combine these two methods effectively. We propose four ways of interactions, that are pipeline framework with tuned orders; knowledge distillation from LLMs to attention masking model; in-context learning with constructed parallel examples. We empirically show these multi-way interactions can improve the baselines in certain perspective of style strength, content preservation and text fluency. Experiments also demonstrate that simply conducting prompting followed by attention masking-based revision can consistently surpass the other systems, including supervised text style transfer systems. On Yelp-clean and Amazon-clean datasets, it improves the previously best mean metric by 0.5 and 3.0 absolute percentages respectively, and achieves new SOTA results.

http://arxiv.org/abs/2402.13647v1

cs.CL, cs.AI

UniGraph: Learning a Cross-Domain Graph Foundation Model From Natural Language

Yufei He, Bryan Hooi

Foundation models like ChatGPT and GPT-4 have revolutionized artificial intelligence, exhibiting remarkable abilities to generalize across a wide array of tasks and applications beyond their initial training objectives. However, when this concept is applied to graph learning, a stark contrast emerges. Graph learning has predominantly focused on single-graph models, tailored to specific tasks or datasets, lacking the ability to transfer learned knowledge to different domains. This limitation stems from the inherent complexity and diversity of graph structures, along with the different feature and label spaces specific to graph data. In this paper, we present our UniGraph framework, designed to train a graph foundation model capable of generalizing to unseen graphs and tasks across diverse domains. Unlike single-graph models that use pre-computed node features of varying dimensions as input, our approach leverages Text-Attributed Graphs (TAGs) for unifying node representations. We propose a cascaded architecture of Language Models (LMs) and Graph Neural Networks (GNNs) as backbone networks with a self-supervised training objective based on Masked Graph Modeling (MGM). We introduce graph instruction tuning using Large Language Models (LLMs) to enable zero-shot prediction ability. Our comprehensive experiments across various graph learning tasks and domains demonstrate the model's effectiveness in self-supervised representation learning on unseen graphs, few-shot in-context transfer, and zero-shot transfer, even surpassing or matching the performance of GNNs that have undergone supervised training on target datasets.

http://arxiv.org/abs/2402.13630v1

cs.LG

FLAME: Self-Supervised Low-Resource Taxonomy Expansion using Large Language Models

Sahil Mishra, Ujjwal Sudev, Tanmoy Chakraborty

Taxonomies represent an arborescence hierarchical structure that establishes relationships among entities to convey knowledge within a specific domain. Each edge in the taxonomy signifies a hypernym-hyponym relationship. Taxonomies find utility in various real-world applications, such as e-commerce search engines and recommendation systems. Consequently, there arises a necessity to enhance these taxonomies over time. However, manually curating taxonomies with neoteric data presents challenges due to limitations in available human resources and the exponential growth of data. Therefore, it becomes imperative to develop automatic taxonomy expansion methods. Traditional supervised taxonomy expansion approaches encounter difficulties stemming from limited resources, primarily due to the small size of existing taxonomies. This scarcity of training data often leads to overfitting. In this paper, we propose FLAME, a novel approach for taxonomy expansion in low-resource environments by harnessing the capabilities of large language models that are trained on extensive real-world knowledge. LLMs help compensate for the scarcity of domain-specific knowledge. Specifically, FLAME leverages prompting in few-shot settings to extract the inherent knowledge within the LLMs, ascertaining the hypernym entities within the taxonomy. Furthermore, it employs reinforcement learning to fine-tune the large language models, resulting in more accurate predictions. Experiments on three real-world benchmark datasets demonstrate the effectiveness of FLAME in real-world scenarios, achieving a remarkable improvement of 18.5% in accuracy and 12.3% in Wu & Palmer metric over eight baselines. Furthermore, we elucidate the strengths and weaknesses of FLAME through an extensive case study, error analysis and ablation studies on the benchmarks.

http://arxiv.org/abs/2402.13623v1

cs.CL, cs.SI

A Comprehensive Study of Multilingual Confidence Estimation on Large Language Models

Boyang Xue, Hongru Wang, Weichao Wang, Rui Wang, Sheng Wang, Zeming Liu, Kam-Fai Wong

The tendency of Large Language Models to generate hallucinations and exhibit overconfidence in predictions raises concerns regarding their reliability. Confidence or uncertainty estimations indicating the extent of trustworthiness of a model's response are essential to developing reliable AI systems. Current research primarily focuses on LLM confidence estimations in English, remaining a void for other widely used languages and impeding the global development of reliable AI applications. This paper introduces a comprehensive investigation of Multi-lingual confidence estimation (MlingConf) on LLMs. First, we introduce an elaborated and expert-checked multilingual QA dataset. Second, we delve into the performance of confidence estimations and examine how these confidence scores can enhance LLM performance through self-refinement across diverse languages. Finally, we propose a cross-lingual confidence estimation method to achieve more precise confidence scores. The experimental results showcase the performance of various confidence estimation methods across different languages as well as present that our proposed cross-lingual confidence estimation technique significantly enhances confidence estimation and outperforms several baseline methods.

http://arxiv.org/abs/2402.13606v1

cs.CL

KorNAT: LLM Alignment Benchmark for Korean Social Values and Common Knowledge

Jiyoung Lee, Minwoo Kim, Seungho Kim, Junghwan Kim, Seunghyun Won, Hwaran Lee, Edward Choi

For Large Language Models (LLMs) to be effectively deployed in a specific country, they must possess an understanding of the nation's culture and basic knowledge. To this end, we introduce National Alignment, which measures an alignment between an LLM and a targeted country from two aspects: social value alignment and common knowledge alignment. Social value alignment evaluates how well the model understands nation-specific social values, while common knowledge alignment examines how well the model captures basic knowledge related to the nation. We constructed KorNAT, the first benchmark that measures national alignment with South Korea. For the social value dataset, we obtained ground truth labels from a large-scale survey involving 6,174 unique Korean participants. For the common knowledge dataset, we constructed samples based on Korean textbooks and GED reference materials. KorNAT contains 4K and 6K multiple-choice questions for social value and common knowledge, respectively. Our dataset creation process is meticulously designed and based on statistical sampling theory and was refined through multiple rounds of human review. The experiment results of seven LLMs reveal that only a few models met our reference score, indicating a potential for further enhancement. KorNAT has received government approval after passing an assessment conducted by a government-affiliated organization dedicated to evaluating dataset quality. Samples and detailed evaluation protocols of our dataset can be found in https://selectstar.ai/ko/papers-national-alignment

http://arxiv.org/abs/2402.13605v4

cs.CL

User-LLM: Efficient LLM Contextualization with User Embeddings

Lin Ning, Luyang Liu, Jiaxing Wu, Neo Wu, Devora Berlowitz, Sushant Prakash, Bradley Green, Shawn O'Banion, Jun Xie

Large language models (LLMs) have revolutionized natural language processing. However, effectively incorporating complex and potentially noisy user interaction data remains a challenge. To address this, we propose User-LLM, a novel framework that leverages user embeddings to contextualize LLMs. These embeddings, distilled from diverse user interactions using self-supervised pretraining, capture latent user preferences and their evolution over time. We integrate these user embeddings with LLMs through cross-attention and soft-prompting, enabling LLMs to dynamically adapt to user context. Our comprehensive experiments on MovieLens, Amazon Review, and Google Local Review datasets demonstrate significant performance gains across various tasks. Notably, our approach outperforms text-prompt-based contextualization on long sequence tasks and tasks that require deep user understanding while being computationally efficient. We further incorporate Perceiver layers to streamline the integration between user encoders and LLMs, reducing computational demands.

http://arxiv.org/abs/2402.13598v1

cs.CL, cs.AI, cs.LG

APTQ: Attention-aware Post-Training Mixed-Precision Quantization for Large Language Models

Ziyi Guan, Hantao Huang, Yupeng Su, Hong Huang, Ngai Wong, Hao Yu

Large Language Models (LLMs) have greatly advanced the natural language processing paradigm. However, the high computational load and huge model sizes pose a grand challenge for deployment on edge devices. To this end, we propose APTQ (Attention-aware Post-Training Mixed-Precision Quantization) for LLMs, which considers not only the second-order information of each layer's weights, but also, for the first time, the nonlinear effect of attention outputs on the entire model. We leverage the Hessian trace as a sensitivity metric for mixed-precision quantization, ensuring an informed precision reduction that retains model performance. Experiments show APTQ surpasses previous quantization methods, achieving an average of 4 bit width a 5.22 perplexity nearly equivalent to full precision in the C4 dataset. In addition, APTQ attains state-of-the-art zero-shot accuracy of 68.24\% and 70.48\% at an average bitwidth of 3.8 in LLaMa-7B and LLaMa-13B, respectively, demonstrating its effectiveness to produce high-quality quantized LLMs.

http://arxiv.org/abs/2402.14866v2

cs.LG, cs.AI, cs.CL

ActiveRAG: Revealing the Treasures of Knowledge via Active Learning

Zhipeng Xu, Zhenghao Liu, Yibin Liu, Chenyan Xiong, Yukun Yan, Shuo Wang, Shi Yu, Zhiyuan Liu, Ge Yu

Retrieval Augmented Generation (RAG) has introduced a new paradigm for Large Language Models (LLMs), aiding in the resolution of knowledge-intensive tasks. However, current RAG models position LLMs as passive knowledge receptors, thereby restricting their capacity for learning and comprehending external knowledge. In this paper, we present ActiveRAG, an innovative RAG framework that shifts from passive knowledge acquisition to an active learning mechanism. This approach utilizes the Knowledge Construction mechanism to develop a deeper understanding of external knowledge by associating it with previously acquired or memorized knowledge. Subsequently, it designs the Cognitive Nexus mechanism to incorporate the outcomes from both chains of thought and knowledge construction, thereby calibrating the intrinsic cognition of LLMs. Our experimental results demonstrate that ActiveRAG surpasses previous RAG models, achieving a 5% improvement on question-answering datasets. All data and codes are available at https://github.com/OpenMatch/ActiveRAG.

http://arxiv.org/abs/2402.13547v1

cs.CL

Test-Driven Development for Code Generation

Noble Saji Mathews, Meiyappan Nagappan

Large language models (LLMs) like GPT4, have shown proficiency in generating code snippets from problem statements. Traditionally software development by humans followed a similar methodology of writing code from problem statements or requirements. However, in the past, there have been several studies that have shown the value of test-driven development (TDD) where humans write tests based on problem statements before the code for the functionality is written. In the context of LLM-based code generation, one obvious benefit of TDD is that the developer then knows for sure if the generated code has passed all the given tests or not. Therefore, in this paper, we want to empirically evaluate the hypothesis: giving the problem statements and tests as input to GPT4 is better than just giving the problem statement as input. To test our hypothesis, we build a framework TGen. In our experiments on the MBPP, HumanEval and CodeChef datasets, we consistently find that including tests solves more programming problems than not including them. Thus we show that TDD is a better development model than just using a problem statement when using GPT4 for code generation tasks.

http://arxiv.org/abs/2402.13521v1

cs.SE, cs.AI

Self-DC: When to retrieve and When to generate? Self Divide-and-Conquer for Compositional Unknown Questions

Hongru Wang, Boyang Xue, Baohang Zhou, Tianhua Zhang, Cunxiang Wang, Guanhua Chen, Huimin Wang, Kam-fai Wong

Retrieve-then-read and generate-then-read are two typical solutions to handle unknown and known questions in open-domain question-answering, while the former retrieves necessary external knowledge and the later prompt the large language models to generate internal known knowledge encoded in the parameters. However, few of previous works consider the compositional unknown questions, which consist of several known or unknown sub-questions. Thus, simple binary classification (known or unknown) becomes sub-optimal and inefficient since it will call external retrieval excessively for each compositional unknown question. To this end, we propose the first Compositional unknown Question-Answering dataset (CuQA), and introduce a Self Divide-and-Conquer (Self-DC) framework to empower LLMs to adaptively call different methods on-demand, resulting in better performance and efficiency. Experimental results on two datasets (CuQA and FreshQA) demonstrate that Self-DC can achieve comparable or even better performance with much more less retrieval times compared with several strong baselines.

http://arxiv.org/abs/2402.13514v1

cs.CL, cs.AI

Leveraging Translation For Optimal Recall: Tailoring LLM Personalization With User Profiles

Karthik Ravichandran, Sarmistha Sarna Gomasta

This paper explores a novel technique for improving recall in cross-language information retrieval (CLIR) systems using iterative query refinement grounded in the user's lexical-semantic space. The proposed methodology combines multi-level translation, semantic embedding-based expansion, and user profile-centered augmentation to address the challenge of matching variance between user queries and relevant documents. Through an initial BM25 retrieval, translation into intermediate languages, embedding lookup of similar terms, and iterative re-ranking, the technique aims to expand the scope of potentially relevant results personalized to the individual user. Comparative experiments on news and Twitter datasets demonstrate superior performance over baseline BM25 ranking for the proposed approach across ROUGE metrics. The translation methodology also showed maintained semantic accuracy through the multi-step process. This personalized CLIR framework paves the path for improved context-aware retrieval attentive to the nuances of user language.

http://arxiv.org/abs/2402.13500v1

cs.IR, cs.CL, F.2.2; I.2.7

GradSafe: Detecting Unsafe Prompts for LLMs via Safety-Critical Gradient Analysis

Yueqi Xie, Minghong Fang, Renjie Pi, Neil Gong

Large Language Models (LLMs) face threats from unsafe prompts. Existing methods for detecting unsafe prompts are primarily online moderation APIs or finetuned LLMs. These strategies, however, often require extensive and resource-intensive data collection and training processes. In this study, we propose GradSafe, which effectively detects unsafe prompts by scrutinizing the gradients of safety-critical parameters in LLMs. Our methodology is grounded in a pivotal observation: the gradients of an LLM's loss for unsafe prompts paired with compliance response exhibit similar patterns on certain safety-critical parameters. In contrast, safe prompts lead to markedly different gradient patterns. Building on this observation, GradSafe analyzes the gradients from prompts (paired with compliance responses) to accurately detect unsafe prompts. We show that GradSafe, applied to Llama-2 without further training, outperforms Llama Guard, despite its extensive finetuning with a large dataset, in detecting unsafe prompts. This superior performance is consistent across both zero-shot and adaptation scenarios, as evidenced by our evaluations on the ToxicChat and XSTest. The source code is available at https://github.com/xyq7/GradSafe.

http://arxiv.org/abs/2402.13494v1

cs.CL, cs.CR

ProPD: Dynamic Token Tree Pruning and Generation for LLM Parallel Decoding

Shuzhang Zhong, Zebin Yang, Meng Li, Ruihao Gong, Runsheng Wang, Ru Huang

Recent advancements in generative large language models (LLMs) have significantly boosted the performance in natural language processing tasks. However, their efficiency is hampered by the inherent limitations in autoregressive token generation. While parallel decoding with token tree verification, e.g., Medusa, has been proposed to improve decoding parallelism and efficiency, it often struggles with maintaining contextual relationships due to its independent token prediction approach and incurs significant verification overhead, especially with large tree sizes and batch processing. In this paper, we propose ProPD, an efficient LLM parallel decoding framework based on dynamic token tree pruning and generation. ProPD features an advanced early pruning mechanism to efficiently eliminate unpromising token sequences to improve verification efficiency. Additionally, it introduces a dynamic token tree generation algorithm to balance the computation and parallelism of the verification phase in real-time and maximize the overall efficiency across different batch sizes, sequence lengths, and tasks, etc. We verify ProPD across a diverse set of datasets, LLMs, and batch sizes and demonstrate ProPD consistently outperforms existing decoding algorithms by 1.1-3.2x.

http://arxiv.org/abs/2402.13485v1

cs.LG, cs.CL

Retrieval-Augmented Data Augmentation for Low-Resource Domain Tasks

Minju Seo, Jinheon Baek, James Thorne, Sung Ju Hwang

Despite large successes of recent language models on diverse tasks, they suffer from severe performance degeneration in low-resource settings with limited training data available. Many existing works tackle this problem by generating synthetic data from the training data and then training models on them, recently using Large Language Models (LLMs). However, in low-resource settings, the amount of seed data samples to use for data augmentation is very small, which makes generated samples suboptimal and less diverse. To tackle this challenge, we propose a novel method that augments training data by incorporating a wealth of examples from other datasets, along with the given training data. Specifically, we first retrieve the relevant instances from other datasets, such as their input-output pairs or contexts, based on their similarities with the given seed data, and then prompt LLMs to generate new samples with the contextual information within and across the original and retrieved samples. This approach can ensure that the generated data is not only relevant but also more diverse than what could be achieved using the limited seed data alone. We validate our proposed Retrieval-Augmented Data Augmentation (RADA) framework on multiple datasets under low-resource settings of training and test-time data augmentation scenarios, on which it outperforms existing LLM-powered data augmentation baselines.

http://arxiv.org/abs/2402.13482v1

cs.CL, cs.AI, cs.LG

LLM Jailbreak Attack versus Defense Techniques -- A Comprehensive Study

Zihao Xu, Yi Liu, Gelei Deng, Yuekang Li, Stjepan Picek

Large Language Models (LLMS) have increasingly become central to generating content with potential societal impacts. Notably, these models have demonstrated capabilities for generating content that could be deemed harmful. To mitigate these risks, researchers have adopted safety training techniques to align model outputs with societal values to curb the generation of malicious content. However, the phenomenon of "jailbreaking", where carefully crafted prompts elicit harmful responses from models, persists as a significant challenge. This research conducts a comprehensive analysis of existing studies on jailbreaking LLMs and their defense techniques. We meticulously investigate nine attack techniques and seven defense techniques applied across three distinct language models: Vicuna, LLama, and GPT-3.5 Turbo. We aim to evaluate the effectiveness of these attack and defense techniques. Our findings reveal that existing white-box attacks underperform compared to universal techniques and that including special tokens in the input significantly affects the likelihood of successful attacks. This research highlights the need to concentrate on the security facets of LLMs. Additionally, we contribute to the field by releasing our datasets and testing framework, aiming to foster further research into LLM security. We believe these contributions will facilitate the exploration of security measures within this domain.

http://arxiv.org/abs/2402.13457v1

cs.CR, cs.AI

Ranking Large Language Models without Ground Truth

Amit Dhurandhar, Rahul Nair, Moninder Singh, Elizabeth Daly, Karthikeyan Natesan Ramamurthy

Evaluation and ranking of large language models (LLMs) has become an important problem with the proliferation of these models and their impact. Evaluation methods either require human responses which are expensive to acquire or use pairs of LLMs to evaluate each other which can be unreliable. In this paper, we provide a novel perspective where, given a dataset of prompts (viz. questions, instructions, etc.) and a set of LLMs, we rank them without access to any ground truth or reference responses. Inspired by real life where both an expert and a knowledgeable person can identify a novice our main idea is to consider triplets of models, where each one of them evaluates the other two, correctly identifying the worst model in the triplet with high probability. We also analyze our idea and provide sufficient conditions for it to succeed. Applying this idea repeatedly, we propose two methods to rank LLMs. In experiments on different generative tasks (summarization, multiple-choice, and dialog), our methods reliably recover close to true rankings without reference data. This points to a viable low-resource mechanism for practical use.

http://arxiv.org/abs/2402.14860v2

cs.CL, cs.AI, cs.LG

Unlocking the `Why' of Buying: Introducing a New Dataset and Benchmark for Purchase Reason and Post-Purchase Experience

Tao Chen, Siqi Zuo, Cheng Li, Mingyang Zhang, Qiaozhu Mei, Michael Bendersky

Explanations are crucial for enhancing user trust and understanding within modern recommendation systems. To build truly explainable systems, we need high-quality datasets that elucidate why users make choices. While previous efforts have focused on extracting users' post-purchase sentiment in reviews, they ignore the reasons behind the decision to buy. In our work, we propose a novel purchase reason explanation task. To this end, we introduce an LLM-based approach to generate a dataset that consists of textual explanations of why real users make certain purchase decisions. We induce LLMs to explicitly distinguish between the reasons behind purchasing a product and the experience after the purchase in a user review. An automated, LLM-driven evaluation, as well as a small scale human evaluation, confirms the effectiveness of our approach to obtaining high-quality, personalized explanations. We benchmark this dataset on two personalized explanation generation tasks. We release the code and prompts to spur further research.

http://arxiv.org/abs/2402.13417v1

cs.IR

Harnessing Large Language Models as Post-hoc Correctors

Zhiqiang Zhong, Kuangyu Zhou, Davide Mottin

As Machine Learning (ML) models grow in size and demand higher-quality training data, the expenses associated with re-training and fine-tuning these models are escalating rapidly. Inspired by recent impressive achievements of Large Language Models (LLMs) in different fields, this paper delves into the question: can LLMs efficiently improve an ML's performance at a minimal cost? We show that, through our proposed training-free framework LlmCorr, an LLM can work as a post-hoc corrector to propose corrections for the predictions of an arbitrary ML model. In particular, we form a contextual knowledge database by incorporating the dataset's label information and the ML model's predictions on the validation dataset. Leveraging the in-context learning capability of LLMs, we ask the LLM to summarise the instances in which the ML model makes mistakes and the correlation between primary predictions and true labels. Following this, the LLM can transfer its acquired knowledge to suggest corrections for the ML model's predictions. Our experimental results on the challenging molecular predictions show that LlmCorr improves the performance of a number of models by up to 39%.

http://arxiv.org/abs/2402.13414v1

cs.LG, cs.CL

EvoGrad: A Dynamic Take on the Winograd Schema Challenge with Human Adversaries

Jing Han Sun, Ali Emami

While Large Language Models (LLMs) excel at the Winograd Schema Challenge (WSC), a coreference resolution task testing common-sense reasoning through pronoun disambiguation, they struggle with instances that feature minor alterations or rewording. To address this, we introduce EvoGrad, an open-source platform that harnesses a human-in-the-loop approach to create a dynamic dataset tailored to such altered WSC instances. Leveraging ChatGPT's capabilities, we expand our task instances from 182 to 3,691, setting a new benchmark for diverse common-sense reasoning datasets. Additionally, we introduce the error depth metric, assessing model stability in dynamic tasks. Our results emphasize the challenge posed by EvoGrad: Even the best performing LLM, GPT-3.5, achieves an accuracy of 65.0% with an average error depth of 7.2, a stark contrast to human performance of 92. 8% accuracy without perturbation errors. This highlights ongoing model limitations and the value of dynamic datasets in uncovering them.

http://arxiv.org/abs/2402.13372v2

cs.CL

ChatEL: Entity Linking with Chatbots

Yifan Ding, Qingkai Zeng, Tim Weninger

Entity Linking (EL) is an essential and challenging task in natural language processing that seeks to link some text representing an entity within a document or sentence with its corresponding entry in a dictionary or knowledge base. Most existing approaches focus on creating elaborate contextual models that look for clues the words surrounding the entity-text to help solve the linking problem. Although these fine-tuned language models tend to work, they can be unwieldy, difficult to train, and do not transfer well to other domains. Fortunately, Large Language Models (LLMs) like GPT provide a highly-advanced solution to the problems inherent in EL models, but simply naive prompts to LLMs do not work well. In the present work, we define ChatEL, which is a three-step framework to prompt LLMs to return accurate results. Overall the ChatEL framework improves the average F1 performance across 10 datasets by more than 2%. Finally, a thorough error analysis shows many instances with the ground truth labels were actually incorrect, and the labels predicted by ChatEL were actually correct. This indicates that the quantitative results presented in this paper may be a conservative estimate of the actual performance. All data and code are available as an open-source package on GitHub at https://github.com/yifding/In_Context_EL.

http://arxiv.org/abs/2402.14858v1

cs.CL, cs.AI

BiMediX: Bilingual Medical Mixture of Experts LLM

Sara Pieri, Sahal Shaji Mullappilly, Fahad Shahbaz Khan, Rao Muhammad Anwer, Salman Khan, Timothy Baldwin, Hisham Cholakkal

In this paper, we introduce BiMediX, the first bilingual medical mixture of experts LLM designed for seamless interaction in both English and Arabic. Our model facilitates a wide range of medical interactions in English and Arabic, including multi-turn chats to inquire about additional details such as patient symptoms and medical history, multiple-choice question answering, and open-ended question answering. We propose a semi-automated English-to-Arabic translation pipeline with human refinement to ensure high-quality translations. We also introduce a comprehensive evaluation benchmark for Arabic medical LLMs. Furthermore, we introduce BiMed1.3M, an extensive Arabic-English bilingual instruction set covering 1.3 Million diverse medical interactions, resulting in over 632 million healthcare specialized tokens for instruction tuning. Our BiMed1.3M dataset includes 250k synthesized multi-turn doctor-patient chats and maintains a 1:2 Arabic-to-English ratio. Our model outperforms state-of-the-art Med42 and Meditron by average absolute gains of 2.5% and 4.1%, respectively, computed across multiple medical evaluation benchmarks in English, while operating at 8-times faster inference. Moreover, our BiMediX outperforms the generic Arabic-English bilingual LLM, Jais-30B, by average absolute gains of 10% on our Arabic medical benchmark and 15% on bilingual evaluations across multiple datasets. Our project page with source code and trained model is available at https://github.com/mbzuai-oryx/BiMediX .

http://arxiv.org/abs/2402.13253v1

cs.CL

Exploring the Frontier of Vision-Language Models: A Survey of Current Methodologies and Future Directions

Akash Ghosh, Arkadeep Acharya, Sriparna Saha, Vinija Jain, Aman Chadha

The advent of Large Language Models (LLMs) has significantly reshaped the trajectory of the AI revolution. Nevertheless, these LLMs exhibit a notable limitation, as they are primarily adept at processing textual information. To address this constraint, researchers have endeavored to integrate visual capabilities with LLMs, resulting in the emergence of Vision-Language Models (VLMs). These advanced models are instrumental in tackling more intricate tasks such as image captioning and visual question answering. In our comprehensive survey paper, we delve into the key advancements within the realm of VLMs. Our classification organizes VLMs into three distinct categories: models dedicated to vision-language understanding, models that process multimodal inputs to generate unimodal (textual) outputs and models that both accept and produce multimodal inputs and outputs.This classification is based on their respective capabilities and functionalities in processing and generating various modalities of data.We meticulously dissect each model, offering an extensive analysis of its foundational architecture, training data sources, as well as its strengths and limitations wherever possible, providing readers with a comprehensive understanding of its essential components. We also analyzed the performance of VLMs in various benchmark datasets. By doing so, we aim to offer a nuanced understanding of the diverse landscape of VLMs. Additionally, we underscore potential avenues for future research in this dynamic domain, anticipating further breakthroughs and advancements.

http://arxiv.org/abs/2404.07214v2

cs.CV, cs.AI, cs.CL

Investigating Cultural Alignment of Large Language Models

Badr AlKhamissi, Muhammad ElNokrashy, Mai AlKhamissi, Mona Diab

The intricate relationship between language and culture has long been a subject of exploration within the realm of linguistic anthropology. Large Language Models (LLMs), promoted as repositories of collective human knowledge, raise a pivotal question: do these models genuinely encapsulate the diverse knowledge adopted by different cultures? Our study reveals that these models demonstrate greater cultural alignment along two dimensions -- firstly, when prompted with the dominant language of a specific culture, and secondly, when pretrained with a refined mixture of languages employed by that culture. We quantify cultural alignment by simulating sociological surveys, comparing model responses to those of actual survey participants as references. Specifically, we replicate a survey conducted in various regions of Egypt and the United States through prompting LLMs with different pretraining data mixtures in both Arabic and English with the personas of the real respondents and the survey questions. Further analysis reveals that misalignment becomes more pronounced for underrepresented personas and for culturally sensitive topics, such as those probing social values. Finally, we introduce Anthropological Prompting, a novel method leveraging anthropological reasoning to enhance cultural alignment. Our study emphasizes the necessity for a more balanced multilingual pretraining dataset to better represent the diversity of human experience and the plurality of different cultures with many implications on the topic of cross-lingual transfer.

http://arxiv.org/abs/2402.13231v1

cs.CL, cs.CY

Smaug: Fixing Failure Modes of Preference Optimisation with DPO-Positive

Arka Pal, Deep Karkhanis, Samuel Dooley, Manley Roberts, Siddartha Naidu, Colin White

Direct Preference Optimisation (DPO) is effective at significantly improving the performance of large language models (LLMs) on downstream tasks such as reasoning, summarisation, and alignment. Using pairs of preferred and dispreferred data, DPO models the \textit{relative} probability of picking one response over another. In this work, first we show theoretically that the standard DPO loss can lead to a \textit{reduction} of the model's likelihood of the preferred examples, as long as the relative probability between the preferred and dispreferred classes increases. We then show empirically that this phenomenon occurs when fine-tuning LLMs on common datasets, especially datasets in which the edit distance between pairs of completions is low. Using these insights, we design DPO-Positive (DPOP), a new loss function and training procedure which avoids this failure mode. Surprisingly, we also find that DPOP significantly outperforms DPO across a wide variety of datasets and downstream tasks, including datasets with high edit distances between completions. By fine-tuning with DPOP, we create and release Smaug-34B and Smaug-72B, which achieve state-of-the-art open-source performance. Notably, Smaug-72B is nearly 2\% better than any other open-source model on the HuggingFace Open LLM Leaderboard and becomes the first open-source LLM to surpass an average accuracy of 80\%.

http://arxiv.org/abs/2402.13228v1

cs.CL, cs.AI, cs.LG

RoCode: A Dataset for Measuring Code Intelligence from Problem Definitions in Romanian

Adrian Cosma, Bogdan Iordache, Paolo Rosso

Recently, large language models (LLMs) have become increasingly powerful and have become capable of solving a plethora of tasks through proper instructions in natural language. However, the vast majority of testing suites assume that the instructions are written in English, the de facto prompting language. Code intelligence and problem solving still remain a difficult task, even for the most advanced LLMs. Currently, there are no datasets to measure the generalization power for code-generation models in a language other than English. In this work, we present RoCode, a competitive programming dataset, consisting of 2,642 problems written in Romanian, 11k solutions in C, C++ and Python and comprehensive testing suites for each problem. The purpose of RoCode is to provide a benchmark for evaluating the code intelligence of language models trained on Romanian / multilingual text as well as a fine-tuning set for pretrained Romanian models. Through our results and review of related works, we argue for the need to develop code models for languages other than English.

http://arxiv.org/abs/2402.13222v1

cs.CL

Softmax Probabilities (Mostly) Predict Large Language Model Correctness on Multiple-Choice Q&A

Benjamin Plaut, Khanh Nguyen, Tu Trinh

Although large language models (LLMs) perform impressively on many tasks, overconfidence remains a problem. We hypothesized that on multiple-choice Q&A tasks, wrong answers would be associated with smaller maximum softmax probabilities (MSPs) compared to correct answers. We comprehensively evaluate this hypothesis on ten open-source LLMs and five datasets, and find strong evidence for our hypothesis among models which perform well on the original Q&A task. For the six LLMs with the best Q&A performance, the AUROC derived from the MSP was better than random chance with p < 10^{-4} in 59/60 instances. Among those six LLMs, the average AUROC ranged from 60% to 69%. Leveraging these findings, we propose a multiple-choice Q&A task with an option to abstain and show that performance can be improved by selectively abstaining based on the MSP of the initial model response. We also run the same experiments with pre-softmax logits instead of softmax probabilities and find similar (but not identical) results.

http://arxiv.org/abs/2402.13213v1

cs.CL, cs.AI, cs.LG

Can Large Language Models be Good Emotional Supporter? Mitigating Preference Bias on Emotional Support Conversation

Dongjin Kang, Sunghwan Kim, Taeyoon Kwon, Seungjun Moon, Hyunsouk Cho, Youngjae Yu, Dongha Lee, Jinyoung Yeo

Emotional Support Conversation (ESC) is a task aimed at alleviating individuals' emotional distress through daily conversation. Given its inherent complexity and non-intuitive nature, ESConv dataset incorporates support strategies to facilitate the generation of appropriate responses. Recently, despite the remarkable conversational ability of large language models (LLMs), previous studies have suggested that they often struggle with providing useful emotional support. Hence, this work initially analyzes the results of LLMs on ESConv, revealing challenges in selecting the correct strategy and a notable preference for a specific strategy. Motivated by these, we explore the impact of the inherent preference in LLMs on providing emotional support, and consequently, we observe that exhibiting high preference for specific strategies hinders effective emotional support, aggravating its robustness in predicting the appropriate strategy. Moreover, we conduct a methodological study to offer insights into the necessary approaches for LLMs to serve as proficient emotional supporters. Our findings emphasize that (1) low preference for specific strategies hinders the progress of emotional support, (2) external assistance helps reduce preference bias, and (3) LLMs alone cannot become good emotional supporters. These insights suggest promising avenues for future research to enhance the emotional intelligence of LLMs.

http://arxiv.org/abs/2402.13211v1

cs.CL, I.2.7

What if LLMs Have Different World Views: Simulating Alien Civilizations with LLM-based Agents

Mingyu Jin, Beichen Wang, Zhaoqian Xue, Suiyuan Zhu, Wenyue Hua, Hua Tang, Kai Mei, Mengnan Du, Yongfeng Zhang

In this study, we introduce "CosmoAgent," an innovative artificial intelligence framework utilizing Large Language Models (LLMs) to simulate complex interactions between human and extraterrestrial civilizations, with a special emphasis on Stephen Hawking's cautionary advice about not sending radio signals haphazardly into the universe. The goal is to assess the feasibility of peaceful coexistence while considering potential risks that could threaten well-intentioned civilizations. Employing mathematical models and state transition matrices, our approach quantitatively evaluates the development trajectories of civilizations, offering insights into future decision-making at critical points of growth and saturation. Furthermore, the paper acknowledges the vast diversity in potential living conditions across the universe, which could foster unique cosmologies, ethical codes, and worldviews among various civilizations. Recognizing the Earth-centric bias inherent in current LLM designs, we propose the novel concept of using LLMs with diverse ethical paradigms and simulating interactions between entities with distinct moral principles. This innovative research provides a new way to understand complex inter-civilizational dynamics, expanding our perspective while pioneering novel strategies for conflict resolution, crucial for preventing interstellar conflicts. We have also released the code and datasets to enable further academic investigation into this interesting area of research. The code is available at https://github.com/agiresearch/AlienAgent.

http://arxiv.org/abs/2402.13184v2

cs.CL

Benchmarking Retrieval-Augmented Generation for Medicine

Guangzhi Xiong, Qiao Jin, Zhiyong Lu, Aidong Zhang

While large language models (LLMs) have achieved state-of-the-art performance on a wide range of medical question answering (QA) tasks, they still face challenges with hallucinations and outdated knowledge. Retrieval-augmented generation (RAG) is a promising solution and has been widely adopted. However, a RAG system can involve multiple flexible components, and there is a lack of best practices regarding the optimal RAG setting for various medical purposes. To systematically evaluate such systems, we propose the Medical Information Retrieval-Augmented Generation Evaluation (MIRAGE), a first-of-its-kind benchmark including 7,663 questions from five medical QA datasets. Using MIRAGE, we conducted large-scale experiments with over 1.8 trillion prompt tokens on 41 combinations of different corpora, retrievers, and backbone LLMs through the MedRAG toolkit introduced in this work. Overall, MedRAG improves the accuracy of six different LLMs by up to 18% over chain-of-thought prompting, elevating the performance of GPT-3.5 and Mixtral to GPT-4-level. Our results show that the combination of various medical corpora and retrievers achieves the best performance. In addition, we discovered a log-linear scaling property and the "lost-in-the-middle" effects in medical RAG. We believe our comprehensive evaluations can serve as practical guidelines for implementing RAG systems for medicine.

http://arxiv.org/abs/2402.13178v2

cs.CL, cs.AI

Defending Jailbreak Prompts via In-Context Adversarial Game

Yujun Zhou, Yufei Han, Haomin Zhuang, Taicheng Guo, Kehan Guo, Zhenwen Liang, Hongyan Bao, Xiangliang Zhang

Large Language Models (LLMs) demonstrate remarkable capabilities across diverse applications. However, concerns regarding their security, particularly the vulnerability to jailbreak attacks, persist. Drawing inspiration from adversarial training in deep learning and LLM agent learning processes, we introduce the In-Context Adversarial Game (ICAG) for defending against jailbreaks without the need for fine-tuning. ICAG leverages agent learning to conduct an adversarial game, aiming to dynamically extend knowledge to defend against jailbreaks. Unlike traditional methods that rely on static datasets, ICAG employs an iterative process to enhance both the defense and attack agents. This continuous improvement process strengthens defenses against newly generated jailbreak prompts. Our empirical studies affirm ICAG's efficacy, where LLMs safeguarded by ICAG exhibit significantly reduced jailbreak success rates across various attack scenarios. Moreover, ICAG demonstrates remarkable transferability to other LLMs, indicating its potential as a versatile defense mechanism.

http://arxiv.org/abs/2402.13148v1

cs.LG, cs.CR

OLViT: Multi-Modal State Tracking via Attention-Based Embeddings for Video-Grounded Dialog

Adnen Abdessaied, Manuel von Hochmeister, Andreas Bulling

We present the Object Language Video Transformer (OLViT) - a novel model for video dialog operating over a multi-modal attention-based dialog state tracker. Existing video dialog models struggle with questions requiring both spatial and temporal localization within videos, long-term temporal reasoning, and accurate object tracking across multiple dialog turns. OLViT addresses these challenges by maintaining a global dialog state based on the output of an Object State Tracker (OST) and a Language State Tracker (LST): while the OST attends to the most important objects within the video, the LST keeps track of the most important linguistic co-references to previous dialog turns. In stark contrast to previous works, our approach is generic by nature and is therefore capable of learning continuous multi-modal dialog state representations of the most relevant objects and rounds. As a result, they can be seamlessly integrated into Large Language Models (LLMs) and offer high flexibility in dealing with different datasets and tasks. Evaluations on the challenging DVD (response classification) and SIMMC 2.1 (response generation) datasets show that OLViT achieves new state-of-the-art performance across both datasets.

http://arxiv.org/abs/2402.13146v1

cs.CV

CIF-Bench: A Chinese Instruction-Following Benchmark for Evaluating the Generalizability of Large Language Models

Yizhi LI, Ge Zhang, Xingwei Qu, Jiali Li, Zhaoqun Li, Zekun Wang, Hao Li, Ruibin Yuan, Yinghao Ma, Kai Zhang, Wangchunshu Zhou, Yiming Liang, Lei Zhang, Lei Ma, Jiajun Zhang, Zuowen Li, Stephen W. Huang, Chenghua Lin, Wenhu Chen, Jie Fu

The advancement of large language models (LLMs) has enhanced the ability to generalize across a wide range of unseen natural language processing (NLP) tasks through instruction-following. Yet, their effectiveness often diminishes in low-resource languages like Chinese, exacerbated by biased evaluations from data leakage, casting doubt on their true generalizability to new linguistic territories. In response, we introduce the Chinese Instruction-Following Benchmark (CIF-Bench), designed to evaluate the zero-shot generalizability of LLMs to the Chinese language. CIF-Bench comprises 150 tasks and 15,000 input-output pairs, developed by native speakers to test complex reasoning and Chinese cultural nuances across 20 categories. To mitigate evaluation bias, we release only half of the dataset publicly, with the remainder kept private, and introduce diversified instructions to minimize score variance, totaling 45,000 data instances. Our evaluation of 28 selected LLMs reveals a noticeable performance gap, with the best model scoring only 52.9%, highlighting the limitations of LLMs in less familiar language and task contexts. This work aims to uncover the current limitations of LLMs in handling Chinese tasks, pushing towards the development of more culturally informed and linguistically diverse models with the released data and benchmark (https://yizhilll.github.io/CIF-Bench/).

http://arxiv.org/abs/2402.13109v1

cs.CL, cs.AI

ELAD: Explanation-Guided Large Language Models Active Distillation

Yifei Zhang, Bo Pan, Chen Ling, Yuntong Hu, Liang Zhao

The deployment and application of Large Language Models (LLMs) is hindered by their memory inefficiency, computational demands, and the high costs of API inferences. Traditional distillation methods, which transfer the capabilities of LLMs to smaller models, often fail to determine whether the knowledge has been sufficiently transferred, potentially resulting in high costs or incomplete distillation. In this paper, we propose an Explanation-Guided LLMs Active Distillation (ELAD) framework that employs an active learning strategy to optimize the balance between annotation costs and model performance. To improve efficient sample selection, we introduce an explanation-guided sample selection method that identifies samples challenging its reasoning by exploiting uncertainties in explanation steps. Additionally, we present a customized LLM-annotated explanation revision technique where the teacher model detects and corrects flaws in the student model's reasoning. Our experiments across various reasoning datasets demonstrate that our framework significantly enhances the efficiency of LLM knowledge distillation.

http://arxiv.org/abs/2402.13098v1

cs.CL, cs.AI