Datasets:

bunkalab

/

arxiv_datasets_llm

like 0

Re-Search for The Truth: Multi-round Retrieval-augmented Large Language Models are Strong Fake News Detectors

Guanghua Li, Wensheng Lu, Wei Zhang, Defu Lian, Kezhong Lu, Rui Mao, Kai Shu, Hao Liao

The proliferation of fake news has had far-reaching implications on politics, the economy, and society at large. While Fake news detection methods have been employed to mitigate this issue, they primarily depend on two essential elements: the quality and relevance of the evidence, and the effectiveness of the verdict prediction mechanism. Traditional methods, which often source information from static repositories like Wikipedia, are limited by outdated or incomplete data, particularly for emerging or rare claims. Large Language Models (LLMs), known for their remarkable reasoning and generative capabilities, introduce a new frontier for fake news detection. However, like traditional methods, LLM-based solutions also grapple with the limitations of stale and long-tail knowledge. Additionally, retrieval-enhanced LLMs frequently struggle with issues such as low-quality evidence retrieval and context length constraints. To address these challenges, we introduce a novel, retrieval-augmented LLMs framework--the first of its kind to automatically and strategically extract key evidence from web sources for claim verification. Employing a multi-round retrieval strategy, our framework ensures the acquisition of sufficient, relevant evidence, thereby enhancing performance. Comprehensive experiments across three real-world datasets validate the framework's superiority over existing methods. Importantly, our model not only delivers accurate verdicts but also offers human-readable explanations to improve result interpretability.

http://arxiv.org/abs/2403.09747v1

cs.CL, cs.AI

Second-Order Information Matters: Revisiting Machine Unlearning for Large Language Models

Kang Gu, Md Rafi Ur Rashid, Najrin Sultana, Shagufta Mehnaz

With the rapid development of Large Language Models (LLMs), we have witnessed intense competition among the major LLM products like ChatGPT, LLaMa, and Gemini. However, various issues (e.g. privacy leakage and copyright violation) of the training corpus still remain underexplored. For example, the Times sued OpenAI and Microsoft for infringing on its copyrights by using millions of its articles for training. From the perspective of LLM practitioners, handling such unintended privacy violations can be challenging. Previous work addressed the ``unlearning" problem of LLMs using gradient information, while they mostly introduced significant overheads like data preprocessing or lacked robustness. In this paper, contrasting with the methods based on first-order information, we revisit the unlearning problem via the perspective of second-order information (Hessian). Our unlearning algorithms, which are inspired by classic Newton update, are not only data-agnostic/model-agnostic but also proven to be robust in terms of utility preservation or privacy guarantee. Through a comprehensive evaluation with four NLP datasets as well as a case study on real-world datasets, our methods consistently show superiority over the first-order methods.

http://arxiv.org/abs/2403.10557v1

cs.LG, cs.AI, cs.CL

Simple and Scalable Strategies to Continually Pre-train Large Language Models

Adam Ibrahim, Benjamin Thérien, Kshitij Gupta, Mats L. Richter, Quentin Anthony, Timothée Lesort, Eugene Belilovsky, Irina Rish

Large language models (LLMs) are routinely pre-trained on billions of tokens, only to start the process over again once new data becomes available. A much more efficient solution is to continually pre-train these models, saving significant compute compared to re-training. However, the distribution shift induced by new data typically results in degraded performance on previous data or poor adaptation to the new data. In this work, we show that a simple and scalable combination of learning rate (LR) re-warming, LR re-decaying, and replay of previous data is sufficient to match the performance of fully re-training from scratch on all available data, as measured by the final loss and the average score on several language model (LM) evaluation benchmarks. Specifically, we show this for a weak but realistic distribution shift between two commonly used LLM pre-training datasets (English$\rightarrow$English) and a stronger distribution shift (English$\rightarrow$German) at the $405$M parameter model scale with large dataset sizes (hundreds of billions of tokens). Selecting the weak but realistic shift for larger-scale experiments, we also find that our continual learning strategies match the re-training baseline for a 10B parameter LLM. Our results demonstrate that LLMs can be successfully updated via simple and scalable continual learning strategies, matching the re-training baseline using only a fraction of the compute. Finally, inspired by previous work, we propose alternatives to the cosine learning rate schedule that help circumvent forgetting induced by LR re-warming and that are not bound to a fixed token budget.

http://arxiv.org/abs/2403.08763v3

cs.LG, cs.AI, cs.CL

Steering LLMs Towards Unbiased Responses: A Causality-Guided Debiasing Framework

Jingling Li, Zeyu Tang, Xiaoyu Liu, Peter Spirtes, Kun Zhang, Liu Leqi, Yang Liu

Large language models (LLMs) can easily generate biased and discriminative responses. As LLMs tap into consequential decision-making (e.g., hiring and healthcare), it is of crucial importance to develop strategies to mitigate these biases. This paper focuses on social bias, tackling the association between demographic information and LLM outputs. We propose a causality-guided debiasing framework that utilizes causal understandings of (1) the data-generating process of the training corpus fed to LLMs, and (2) the internal reasoning process of LLM inference, to guide the design of prompts for debiasing LLM outputs through selection mechanisms. Our framework unifies existing de-biasing prompting approaches such as inhibitive instructions and in-context contrastive examples, and sheds light on new ways of debiasing by encouraging bias-free reasoning. Our strong empirical performance on real-world datasets demonstrates that our framework provides principled guidelines on debiasing LLM outputs even with only the black-box access.

http://arxiv.org/abs/2403.08743v1

cs.CL, cs.AI, cs.LG

Strengthening Multimodal Large Language Model with Bootstrapped Preference Optimization

Renjie Pi, Tianyang Han, Wei Xiong, Jipeng Zhang, Runtao Liu, Rui Pan, Tong Zhang

Multimodal Large Language Models (MLLMs) excel in generating responses based on visual inputs. However, they often suffer from a bias towards generating responses similar to their pretraining corpus, overshadowing the importance of visual information. We treat this bias as a "preference" for pretraining statistics, which hinders the model's grounding in visual input. To mitigate this issue, we propose Bootstrapped Preference Optimization (BPO), which conducts preference learning with datasets containing negative responses bootstrapped from the model itself. Specifically, we propose the following two strategies: 1) using distorted image inputs to the MLLM for eliciting responses that contain signified pretraining bias; 2) leveraging text-based LLM to explicitly inject erroneous but common elements into the original response. Those undesirable responses are paired with original annotated responses from the datasets to construct the preference dataset, which is subsequently utilized to perform preference learning. Our approach effectively suppresses pretrained LLM bias, enabling enhanced grounding in visual inputs. Extensive experimentation demonstrates significant performance improvements across multiple benchmarks, advancing the state-of-the-art in multimodal conversational systems.

http://arxiv.org/abs/2403.08730v2

cs.CL, cs.CV

Review of Generative AI Methods in Cybersecurity

Yagmur Yigit, William J Buchanan, Madjid G Tehrani, Leandros Maglaras

Over the last decade, Artificial Intelligence (AI) has become increasingly popular, especially with the use of chatbots such as ChatGPT, Gemini, and DALL-E. With this rise, large language models (LLMs) and Generative AI (GenAI) have also become more prevalent in everyday use. These advancements strengthen cybersecurity's defensive posture and open up new attack avenues for adversaries as well. This paper provides a comprehensive overview of the current state-of-the-art deployments of GenAI, covering assaults, jailbreaking, and applications of prompt injection and reverse psychology. This paper also provides the various applications of GenAI in cybercrimes, such as automated hacking, phishing emails, social engineering, reverse cryptography, creating attack payloads, and creating malware. GenAI can significantly improve the automation of defensive cyber security processes through strategies such as dataset construction, safe code development, threat intelligence, defensive measures, reporting, and cyberattack detection. In this study, we suggest that future research should focus on developing robust ethical norms and innovative defense mechanisms to address the current issues that GenAI creates and to also further encourage an impartial approach to its future application in cybersecurity. Moreover, we underscore the importance of interdisciplinary approaches further to bridge the gap between scientific developments and ethical considerations.

http://arxiv.org/abs/2403.08701v2

cs.CR

TeaMs-RL: Teaching LLMs to Teach Themselves Better Instructions via Reinforcement Learning

Shangding Gu, Alois Knoll, Ming Jin

The development of Large Language Models (LLMs) often confronts challenges stemming from the heavy reliance on human annotators in the reinforcement learning with human feedback (RLHF) framework, or the frequent and costly external queries tied to the self-instruct paradigm. In this work, we pivot to Reinforcement Learning (RL) -- but with a twist. Diverging from the typical RLHF, which refines LLMs following instruction data training, we use RL to directly generate the foundational instruction dataset that alone suffices for fine-tuning. Our method, TeaMs-RL, uses a suite of textual operations and rules, prioritizing the diversification of training datasets. It facilitates the generation of high-quality data without excessive reliance on external advanced models, paving the way for a single fine-tuning step and negating the need for subsequent RLHF stages. Our findings highlight key advantages of our approach: reduced need for human involvement and fewer model queries (only $5.73\%$ of WizardLM's total), along with enhanced capabilities of LLMs in crafting and comprehending complex instructions compared to strong baselines, and substantially improved model privacy protection.

http://arxiv.org/abs/2403.08694v2

cs.CL

Zero-shot and Few-shot Generation Strategies for Artificial Clinical Records

Erlend Frayling, Jake Lever, Graham McDonald

The challenge of accessing historical patient data for clinical research, while adhering to privacy regulations, is a significant obstacle in medical science. An innovative approach to circumvent this issue involves utilising synthetic medical records that mirror real patient data without compromising individual privacy. The creation of these synthetic datasets, particularly without using actual patient data to train Large Language Models (LLMs), presents a novel solution as gaining access to sensitive patient information to train models is also a challenge. This study assesses the capability of the Llama 2 LLM to create synthetic medical records that accurately reflect real patient information, employing zero-shot and few-shot prompting strategies for comparison against fine-tuned methodologies that do require sensitive patient data during training. We focus on generating synthetic narratives for the History of Present Illness section, utilising data from the MIMIC-IV dataset for comparison. In this work introduce a novel prompting technique that leverages a chain-of-thought approach, enhancing the model's ability to generate more accurate and contextually relevant medical narratives without prior fine-tuning. Our findings suggest that this chain-of-thought prompted approach allows the zero-shot model to achieve results on par with those of fine-tuned models, based on Rouge metrics evaluation.

http://arxiv.org/abs/2403.08664v2

cs.CL, cs.LG

Distilling Named Entity Recognition Models for Endangered Species from Large Language Models

Jesse Atuhurra, Seiveright Cargill Dujohn, Hidetaka Kamigaito, Hiroyuki Shindo, Taro Watanabe

Natural language processing (NLP) practitioners are leveraging large language models (LLM) to create structured datasets from semi-structured and unstructured data sources such as patents, papers, and theses, without having domain-specific knowledge. At the same time, ecological experts are searching for a variety of means to preserve biodiversity. To contribute to these efforts, we focused on endangered species and through in-context learning, we distilled knowledge from GPT-4. In effect, we created datasets for both named entity recognition (NER) and relation extraction (RE) via a two-stage process: 1) we generated synthetic data from GPT-4 of four classes of endangered species, 2) humans verified the factual accuracy of the synthetic data, resulting in gold data. Eventually, our novel dataset contains a total of 3.6K sentences, evenly divided between 1.8K NER and 1.8K RE sentences. The constructed dataset was then used to fine-tune both general BERT and domain-specific BERT variants, completing the knowledge distillation process from GPT-4 to BERT, because GPT-4 is resource intensive. Experiments show that our knowledge transfer approach is effective at creating a NER model suitable for detecting endangered species from texts.

http://arxiv.org/abs/2403.15430v1

cs.CL

MedInsight: A Multi-Source Context Augmentation Framework for Generating Patient-Centric Medical Responses using Large Language Models

Subash Neupane, Shaswata Mitra, Sudip Mittal, Noorbakhsh Amiri Golilarz, Shahram Rahimi, Amin Amirlatifi

Large Language Models (LLMs) have shown impressive capabilities in generating human-like responses. However, their lack of domain-specific knowledge limits their applicability in healthcare settings, where contextual and comprehensive responses are vital. To address this challenge and enable the generation of patient-centric responses that are contextually relevant and comprehensive, we propose MedInsight:a novel retrieval augmented framework that augments LLM inputs (prompts) with relevant background information from multiple sources. MedInsight extracts pertinent details from the patient's medical record or consultation transcript. It then integrates information from authoritative medical textbooks and curated web resources based on the patient's health history and condition. By constructing an augmented context combining the patient's record with relevant medical knowledge, MedInsight generates enriched, patient-specific responses tailored for healthcare applications such as diagnosis, treatment recommendations, or patient education. Experiments on the MTSamples dataset validate MedInsight's effectiveness in generating contextually appropriate medical responses. Quantitative evaluation using the Ragas metric and TruLens for answer similarity and answer correctness demonstrates the model's efficacy. Furthermore, human evaluation studies involving Subject Matter Expert (SMEs) confirm MedInsight's utility, with moderate inter-rater agreement on the relevance and correctness of the generated responses.

http://arxiv.org/abs/2403.08607v1

cs.CL, cs.AI

Call Me When Necessary: LLMs can Efficiently and Faithfully Reason over Structured Environments

Sitao Cheng, Ziyuan Zhuang, Yong Xu, Fangkai Yang, Chaoyun Zhang, Xiaoting Qin, Xiang Huang, Ling Chen, Qingwei Lin, Dongmei Zhang, Saravan Rajmohan, Qi Zhang

Large Language Models (LLMs) have shown potential in reasoning over structured environments, e.g., knowledge graph and table. Such tasks typically require multi-hop reasoning, i.e., match natural language utterance with instances in the environment. Previous methods leverage LLMs to incrementally build a reasoning path, where the LLMs either invoke tools or pick up schemas by step-by-step interacting with the environment. We propose Reasoning-Path-Editing (Readi), a novel framework where LLMs can efficiently and faithfully reason over structured environments. In Readi, LLMs initially generate a reasoning path given a query, and edit the path only when necessary. We instantiate the path on structured environments and provide feedback to edit the path if anything goes wrong. Experimental results on three KGQA datasets and two TableQA datasets show the effectiveness of Readi, significantly surpassing all LLM-based methods (by 9.1% on WebQSP, 12.4% on MQA-3H and 10.9% on WTQ), comparable with state-of-the-art fine-tuned methods (67% on CWQ and 74.7% on WebQSP) and substantially boosting the vanilla LLMs (by 14.9% on CWQ). Our code will be available upon publication.

http://arxiv.org/abs/2403.08593v1

cs.CL, cs.AI

Rich Semantic Knowledge Enhanced Large Language Models for Few-shot Chinese Spell Checking

Ming Dong, Yujing Chen, Miao Zhang, Hao Sun, Tingting He

Chinese Spell Checking (CSC) is a widely used technology, which plays a vital role in speech to text (STT) and optical character recognition (OCR). Most of the existing CSC approaches relying on BERT architecture achieve excellent performance. However, limited by the scale of the foundation model, BERT-based method does not work well in few-shot scenarios, showing certain limitations in practical applications. In this paper, we explore using an in-context learning method named RS-LLM (Rich Semantic based LLMs) to introduce large language models (LLMs) as the foundation model. Besides, we study the impact of introducing various Chinese rich semantic information in our framework. We found that by introducing a small number of specific Chinese rich semantic structures, LLMs achieve better performance than the BERT-based model on few-shot CSC task. Furthermore, we conduct experiments on multiple datasets, and the experimental results verified the superiority of our proposed framework.

http://arxiv.org/abs/2403.08492v1

cs.CL

Search-based Optimisation of LLM Learning Shots for Story Point Estimation

Vali Tawosi, Salwa Alamir, Xiaomo Liu

One of the ways Large Language Models (LLMs) are used to perform machine learning tasks is to provide them with a few examples before asking them to produce a prediction. This is a meta-learning process known as few-shot learning. In this paper, we use available Search-Based methods to optimise the number and combination of examples that can improve an LLM's estimation performance, when it is used to estimate story points for new agile tasks. Our preliminary results show that our SBSE technique improves the estimation performance of the LLM by 59.34% on average (in terms of mean absolute error of the estimation) over three datasets against a zero-shot setting.

http://arxiv.org/abs/2403.08430v1

cs.SE, cs.AI

Software Vulnerability and Functionality Assessment using LLMs

Rasmus Ingemann Tuffveson Jensen, Vali Tawosi, Salwa Alamir

While code review is central to the software development process, it can be tedious and expensive to carry out. In this paper, we investigate whether and how Large Language Models (LLMs) can aid with code reviews. Our investigation focuses on two tasks that we argue are fundamental to good reviews: (i) flagging code with security vulnerabilities and (ii) performing software functionality validation, i.e., ensuring that code meets its intended functionality. To test performance on both tasks, we use zero-shot and chain-of-thought prompting to obtain final ``approve or reject'' recommendations. As data, we employ seminal code generation datasets (HumanEval and MBPP) along with expert-written code snippets with security vulnerabilities from the Common Weakness Enumeration (CWE). Our experiments consider a mixture of three proprietary models from OpenAI and smaller open-source LLMs. We find that the former outperforms the latter by a large margin. Motivated by promising results, we finally ask our models to provide detailed descriptions of security vulnerabilities. Results show that 36.7% of LLM-generated descriptions can be associated with true CWE vulnerabilities.

http://arxiv.org/abs/2403.08429v1

cs.SE, cs.AI

Mastering Text, Code and Math Simultaneously via Fusing Highly Specialized Language Models

Ning Ding, Yulin Chen, Ganqu Cui, Xingtai Lv, Weilin Zhao, Ruobing Xie, Bowen Zhou, Zhiyuan Liu, Maosong Sun

Underlying data distributions of natural language, programming code, and mathematical symbols vary vastly, presenting a complex challenge for large language models (LLMs) that strive to achieve high performance across all three domains simultaneously. Achieving a very high level of proficiency for an LLM within a specific domain often requires extensive training with relevant corpora, which is typically accompanied by a sacrifice in performance in other domains. In this paper, we propose to fuse models that are already highly-specialized directly. The proposed fusing framework, UltraFuser, consists of three distinct specialists that are already sufficiently trained on language, coding, and mathematics. A token-level gating mechanism is introduced to blend the specialists' outputs. A two-stage training strategy accompanied by balanced sampling is designed to ensure stability. To effectively train the fused model, we further construct a high-quality supervised instruction tuning dataset, UltraChat 2, which includes text, code, and mathematical content. This dataset comprises approximately 300,000 instructions and covers a wide range of topics in each domain. Experiments show that our model could simultaneously achieve mastery of the three crucial domains.

http://arxiv.org/abs/2403.08281v4

cs.CL, cs.AI

RECIPE4U: Student-ChatGPT Interaction Dataset in EFL Writing Education

Jieun Han, Haneul Yoo, Junho Myung, Minsun Kim, Tak Yeon Lee, So-Yeon Ahn, Alice Oh

The integration of generative AI in education is expanding, yet empirical analyses of large-scale and real-world interactions between students and AI systems still remain limited. Addressing this gap, we present RECIPE4U (RECIPE for University), a dataset sourced from a semester-long experiment with 212 college students in English as Foreign Language (EFL) writing courses. During the study, students engaged in dialogues with ChatGPT to revise their essays. RECIPE4U includes comprehensive records of these interactions, including conversation logs, students' intent, students' self-rated satisfaction, and students' essay edit histories. In particular, we annotate the students' utterances in RECIPE4U with 13 intention labels based on our coding schemes. We establish baseline results for two subtasks in task-oriented dialogue systems within educational contexts: intent detection and satisfaction estimation. As a foundational step, we explore student-ChatGPT interaction patterns through RECIPE4U and analyze them by focusing on students' dialogue, essay data statistics, and students' essay edits. We further illustrate potential applications of RECIPE4U dataset for enhancing the incorporation of LLMs in educational frameworks. RECIPE4U is publicly available at https://zeunie.github.io/RECIPE4U/.

http://arxiv.org/abs/2403.08272v1

cs.CL

TINA: Think, Interaction, and Action Framework for Zero-Shot Vision Language Navigation

Dingbang Li, Wenzhou Chen, Xin Lin

Zero-shot navigation is a critical challenge in Vision-Language Navigation (VLN) tasks, where the ability to adapt to unfamiliar instructions and to act in unknown environments is essential. Existing supervised learning-based models, trained using annotated data through reinforcement learning, exhibit limitations in generalization capabilities. Large Language Models (LLMs), with their extensive knowledge and emergent reasoning abilities, present a potential pathway for achieving zero-shot navigation. This paper presents a VLN agent based on LLMs, exploring approaches to the zero-shot navigation problem. To compensate for the shortcomings of LLMs in environmental perception, we propose the Thinking, Interacting, and Action (TINA) framework. TINA enables the agent to scrutinize perceptual information and autonomously query key clues within the environment through an introduced question-answering module, thereby aligning instructions with specific perceptual data. The navigation agent's perceptual abilities are enhanced through the TINA framework, while the explicit thought and query processes also improve the navigational procedure's explainability and transparency. We evaluate the performance of our method on the Room-to-Room dataset. The experiment results indicate that our approach improves the navigation performance of LLM-based agents. Our approach also outperformed some supervised learning-based methods, highlighting its efficacy in zero-shot navigation.

http://arxiv.org/abs/2403.08833v1

cs.CV, cs.AI

Empowering Robotics with Large Language Models: osmAG Map Comprehension with LLMs

Fujing Xie, Sören Schwertfeger

Recently, Large Language Models (LLMs) have demonstrated great potential in robotic applications by providing essential general knowledge for situations that can not be pre-programmed beforehand. Generally speaking, mobile robots need to understand maps to execute tasks such as localization or navigation. In this letter, we address the problem of enabling LLMs to comprehend Area Graph, a text-based map representation, in order to enhance their applicability in the field of mobile robotics. Area Graph is a hierarchical, topometric semantic map representation utilizing polygons to demark areas such as rooms, corridors or buildings. In contrast to commonly used map representations, such as occupancy grid maps or point clouds, osmAG (Area Graph in OpensStreetMap format) is stored in a XML textual format naturally readable by LLMs. Furthermore, conventional robotic algorithms such as localization and path planning are compatible with osmAG, facilitating this map representation comprehensible by LLMs, traditional robotic algorithms and humans. Our experiments show that with a proper map representation, LLMs possess the capability to understand maps and answer queries based on that understanding. Following simple fine-tuning of LLaMA2 models, it surpassed ChatGPT-3.5 in tasks involving topology and hierarchy understanding. Our dataset, dataset generation code, fine-tuned LoRA adapters can be accessed at https://github.com/xiefujing/LLM-osmAG-Comprehension.

http://arxiv.org/abs/2403.08228v1

cs.RO

MoleculeQA: A Dataset to Evaluate Factual Accuracy in Molecular Comprehension

Xingyu Lu, He Cao, Zijing Liu, Shengyuan Bai, Leqing Chen, Yuan Yao, Hai-Tao Zheng, Yu Li

Large language models are playing an increasingly significant role in molecular research, yet existing models often generate erroneous information, posing challenges to accurate molecular comprehension. Traditional evaluation metrics for generated content fail to assess a model's accuracy in molecular understanding. To rectify the absence of factual evaluation, we present MoleculeQA, a novel question answering (QA) dataset which possesses 62K QA pairs over 23K molecules. Each QA pair, composed of a manual question, a positive option and three negative options, has consistent semantics with a molecular description from authoritative molecular corpus. MoleculeQA is not only the first benchmark for molecular factual bias evaluation but also the largest QA dataset for molecular research. A comprehensive evaluation on MoleculeQA for existing molecular LLMs exposes their deficiencies in specific areas and pinpoints several particularly crucial factors for molecular understanding.

http://arxiv.org/abs/2403.08192v1

cs.CL, q-bio.BM

CHAI: Clustered Head Attention for Efficient LLM Inference

Saurabh Agarwal, Bilge Acun, Basil Hosmer, Mostafa Elhoushi, Yejin Lee, Shivaram Venkataraman, Dimitris Papailiopoulos, Carole-Jean Wu

Large Language Models (LLMs) with hundreds of billions of parameters have transformed the field of machine learning. However, serving these models at inference time is both compute and memory intensive, where a single request can require multiple GPUs and tens of Gigabytes of memory. Multi-Head Attention is one of the key components of LLMs, which can account for over 50% of LLMs memory and compute requirement. We observe that there is a high amount of redundancy across heads on which tokens they pay attention to. Based on this insight, we propose Clustered Head Attention (CHAI). CHAI combines heads with a high amount of correlation for self-attention at runtime, thus reducing both memory and compute. In our experiments, we show that CHAI is able to reduce the memory requirements for storing K,V cache by up to 21.4% and inference time latency by up to 1.73x without any fine-tuning required. CHAI achieves this with a maximum 3.2% deviation in accuracy across 3 different models (i.e. OPT-66B, LLAMA-7B, LLAMA-33B) and 5 different evaluation datasets.

http://arxiv.org/abs/2403.08058v2

cs.LG, cs.CL

Pix2Pix-OnTheFly: Leveraging LLMs for Instruction-Guided Image Editing

Rodrigo Santos, João Silva, António Branco

The combination of language processing and image processing keeps attracting increased interest given recent impressive advances that leverage the combined strengths of both domains of research. Among these advances, the task of editing an image on the basis solely of a natural language instruction stands out as a most challenging endeavour. While recent approaches for this task resort, in one way or other, to some form of preliminary preparation, training or fine-tuning, this paper explores a novel approach: We propose a preparation-free method that permits instruction-guided image editing on the fly. This approach is organized along three steps properly orchestrated that resort to image captioning and DDIM inversion, followed by obtaining the edit direction embedding, followed by image editing proper. While dispensing with preliminary preparation, our approach demonstrates to be effective and competitive, outperforming recent, state of the art models for this task when evaluated on the MAGICBRUSH dataset.

http://arxiv.org/abs/2403.08004v1

cs.CL, cs.AI, cs.CV

Beyond Text: Frozen Large Language Models in Visual Signal Comprehension

Lei Zhu, Fangyun Wei, Yanye Lu

In this work, we investigate the potential of a large language model (LLM) to directly comprehend visual signals without the necessity of fine-tuning on multi-modal datasets. The foundational concept of our method views an image as a linguistic entity, and translates it to a set of discrete words derived from the LLM's vocabulary. To achieve this, we present the Vision-to-Language Tokenizer, abbreviated as V2T Tokenizer, which transforms an image into a ``foreign language'' with the combined aid of an encoder-decoder, the LLM vocabulary, and a CLIP model. With this innovative image encoding, the LLM gains the ability not only for visual comprehension but also for image denoising and restoration in an auto-regressive fashion-crucially, without any fine-tuning. We undertake rigorous experiments to validate our method, encompassing understanding tasks like image recognition, image captioning, and visual question answering, as well as image denoising tasks like inpainting, outpainting, deblurring, and shift restoration. Code and models are available at https://github.com/zh460045050/V2L-Tokenizer.

http://arxiv.org/abs/2403.07874v1

cs.CV

Fine-tuning Large Language Models with Sequential Instructions

Hanxu Hu, Pinzhen Chen, Edoardo M. Ponti

Large language models (LLMs) struggle to follow a sequence of instructions in a single query as they may ignore or misinterpret part of it. This impairs their performance in complex problems whose solution requires multiple intermediate steps, such as multilingual (translate then answer) and multimodal (caption then answer) tasks. We empirically verify this with open-source LLMs as large as LLaMA-2 70B and Mixtral-8x7B. Targeting the scarcity of sequential instructions in present-day data, we propose sequential instruction tuning, a simple yet effective strategy to automatically augment instruction tuning data and equip LLMs with the ability to execute multiple sequential instructions. After exploring interleaving instructions in existing datasets, such as Alpaca, with a wide range of intermediate tasks, we find that sequential instruction-tuned models consistently outperform the conventional instruction-tuned baselines in downstream tasks involving reasoning, multilingual, and multimodal abilities. To shed further light on our technique, we analyse how adversarial intermediate texts, unseen tasks, prompt verbalization, number of tasks, and prompt length affect SIT. We hope that this method will open new research avenues on instruction tuning for complex tasks.

http://arxiv.org/abs/2403.07794v1

cs.CL

Synth$^2$: Boosting Visual-Language Models with Synthetic Captions and Image Embeddings

Sahand Sharifzadeh, Christos Kaplanis, Shreya Pathak, Dharshan Kumaran, Anastasija Ilic, Jovana Mitrovic, Charles Blundell, Andrea Banino

The creation of high-quality human-labeled image-caption datasets presents a significant bottleneck in the development of Visual-Language Models (VLMs). We propose a novel approach that leverages the strengths of Large Language Models (LLMs) and image generation models to create synthetic image-text pairs for efficient and effective VLM training. Our method employs pretraining a text-to-image model to synthesize image embeddings starting from captions generated by an LLM. These synthetic pairs are then used to train a VLM. Extensive experiments demonstrate that the VLM trained with synthetic data exhibits comparable performance on image captioning, while requiring a fraction of the data used by models trained solely on human-annotated data. In particular, we outperform the baseline by 17% through augmentation with a synthetic dataset. Furthermore, we show that synthesizing in the image embedding space is 25% faster than in the pixel space. This research introduces a promising technique for generating large-scale, customizable image datasets, leading to enhanced VLM performance and wider applicability across various domains, all with improved data efficiency and resource utilization.

http://arxiv.org/abs/2403.07750v1

cs.CV, cs.AI

FineMath: A Fine-Grained Mathematical Evaluation Benchmark for Chinese Large Language Models

Yan Liu, Renren Jin, Lin Shi, Zheng Yao, Deyi Xiong

To thoroughly assess the mathematical reasoning abilities of Large Language Models (LLMs), we need to carefully curate evaluation datasets covering diverse mathematical concepts and mathematical problems at different difficulty levels. In pursuit of this objective, we propose FineMath in this paper, a fine-grained mathematical evaluation benchmark dataset for assessing Chinese LLMs. FineMath is created to cover the major key mathematical concepts taught in elementary school math, which are further divided into 17 categories of math word problems, enabling in-depth analysis of mathematical reasoning abilities of LLMs. All the 17 categories of math word problems are manually annotated with their difficulty levels according to the number of reasoning steps required to solve these problems. We conduct extensive experiments on a wide range of LLMs on FineMath and find that there is still considerable room for improvements in terms of mathematical reasoning capability of Chinese LLMs. We also carry out an in-depth analysis on the evaluation process and methods that have been overlooked previously. These two factors significantly influence the model results and our understanding of their mathematical reasoning capabilities. The dataset will be publicly available soon.

http://arxiv.org/abs/2403.07747v1

cs.CL, cs.AI

KnowCoder: Coding Structured Knowledge into LLMs for Universal Information Extraction

Zixuan Li, Yutao Zeng, Yuxin Zuo, Weicheng Ren, Wenxuan Liu, Miao Su, Yucan Guo, Yantao Liu, Xiang Li, Zhilei Hu, Long Bai, Wei Li, Yidan Liu, Pan Yang, Xiaolong Jin, Jiafeng Guo, Xueqi Cheng

In this paper, we propose KnowCoder, a Large Language Model (LLM) to conduct Universal Information Extraction (UIE) via code generation. KnowCoder aims to develop a kind of unified schema representation that LLMs can easily understand and an effective learning framework that encourages LLMs to follow schemas and extract structured knowledge accurately. To achieve these, KnowCoder introduces a code-style schema representation method to uniformly transform different schemas into Python classes, with which complex schema information, such as constraints among tasks in UIE, can be captured in an LLM-friendly manner. We further construct a code-style schema library covering over $\textbf{30,000}$ types of knowledge, which is the largest one for UIE, to the best of our knowledge. To ease the learning process of LLMs, KnowCoder contains a two-phase learning framework that enhances its schema understanding ability via code pretraining and its schema following ability via instruction tuning. After code pretraining on around $1.5$B automatically constructed data, KnowCoder already attains remarkable generalization ability and achieves relative improvements by $\textbf{49.8%}$ F1, compared to LLaMA2, under the few-shot setting. After instruction tuning, KnowCoder further exhibits strong generalization ability on unseen schemas and achieves up to $\textbf{12.5%}$ and $\textbf{21.9%}$, compared to sota baselines, under the zero-shot setting and the low resource setting, respectively. Additionally, based on our unified schema representations, various human-annotated datasets can simultaneously be utilized to refine KnowCoder, which achieves significant improvements up to $\textbf{7.5%}$ under the supervised setting.

http://arxiv.org/abs/2403.07969v2

cs.LG, cs.AI

LLMvsSmall Model? Large Language Model Based Text Augmentation Enhanced Personality Detection Model

Linmei Hu, Hongyu He, Duokang Wang, Ziwang Zhao, Yingxia Shao, Liqiang Nie

Personality detection aims to detect one's personality traits underlying in social media posts. One challenge of this task is the scarcity of ground-truth personality traits which are collected from self-report questionnaires. Most existing methods learn post features directly by fine-tuning the pre-trained language models under the supervision of limited personality labels. This leads to inferior quality of post features and consequently affects the performance. In addition, they treat personality traits as one-hot classification labels, overlooking the semantic information within them. In this paper, we propose a large language model (LLM) based text augmentation enhanced personality detection model, which distills the LLM's knowledge to enhance the small model for personality detection, even when the LLM fails in this task. Specifically, we enable LLM to generate post analyses (augmentations) from the aspects of semantic, sentiment, and linguistic, which are critical for personality detection. By using contrastive learning to pull them together in the embedding space, the post encoder can better capture the psycho-linguistic information within the post representations, thus improving personality detection. Furthermore, we utilize the LLM to enrich the information of personality labels for enhancing the detection performance. Experimental results on the benchmark datasets demonstrate that our model outperforms the state-of-the-art methods on personality detection.

http://arxiv.org/abs/2403.07581v1

cs.CL

MoAI: Mixture of All Intelligence for Large Language and Vision Models

Byung-Kwan Lee, Beomchan Park, Chae Won Kim, Yong Man Ro

The rise of large language models (LLMs) and instruction tuning has led to the current trend of instruction-tuned large language and vision models (LLVMs). This trend involves either meticulously curating numerous instruction tuning datasets tailored to specific objectives or enlarging LLVMs to manage vast amounts of vision language (VL) data. However, current LLVMs have disregarded the detailed and comprehensive real-world scene understanding available from specialized computer vision (CV) models in visual perception tasks such as segmentation, detection, scene graph generation (SGG), and optical character recognition (OCR). Instead, the existing LLVMs rely mainly on the large capacity and emergent capabilities of their LLM backbones. Therefore, we present a new LLVM, Mixture of All Intelligence (MoAI), which leverages auxiliary visual information obtained from the outputs of external segmentation, detection, SGG, and OCR models. MoAI operates through two newly introduced modules: MoAI-Compressor and MoAI-Mixer. After verbalizing the outputs of the external CV models, the MoAI-Compressor aligns and condenses them to efficiently use relevant auxiliary visual information for VL tasks. MoAI-Mixer then blends three types of intelligence (1) visual features, (2) auxiliary features from the external CV models, and (3) language features by utilizing the concept of Mixture of Experts. Through this integration, MoAI significantly outperforms both open-source and closed-source LLVMs in numerous zero-shot VL tasks, particularly those related to real-world scene understanding such as object existence, positions, relations, and OCR without enlarging the model size or curating extra visual instruction tuning datasets.

http://arxiv.org/abs/2403.07508v1

cs.CV

SmallToLarge (S2L): Scalable Data Selection for Fine-tuning Large Language Models by Summarizing Training Trajectories of Small Models

Yu Yang, Siddhartha Mishra, Jeffrey N Chiang, Baharan Mirzasoleiman

Despite the effectiveness of data selection for large language models (LLMs) during pretraining and instruction fine-tuning phases, improving data efficiency in supervised fine-tuning (SFT) for specialized domains poses significant challenges due to the complexity of fine-tuning data. To bridge this gap, we introduce an effective and scalable data selection method for SFT, SmallToLarge (S2L), which leverages training trajectories from small models to guide the data selection for larger models. We demonstrate through extensive experiments that S2L significantly improves data efficiency in SFT for mathematical problem-solving, reducing the training data to just 11% of the original MathInstruct dataset (Yue et al., 2023) to match full dataset performance while outperforming state-of-the-art data selection algorithms by an average of 4.7% across 6 in- and out-domain evaluation datasets. Remarkably, selecting only 50K data for SFT, S2L achieves a 32.7% accuracy on the most challenging MATH (Hendrycks et al., 2021) benchmark, improving Phi-2 (Li et al., 2023b) by 16.6%. In clinical text summarization on the MIMIC-III dataset (Johnson et al., 2016), S2L again outperforms training on the full dataset using only 50% of the data. Notably, S2L can perform data selection using a reference model 40x smaller than the target model, proportionally reducing the cost of data selection.

http://arxiv.org/abs/2403.07384v1

cs.CL, cs.AI, cs.LG

SVD-LLM: Truncation-aware Singular Value Decomposition for Large Language Model Compression

Xin Wang, Yu Zheng, Zhongwei Wan, Mi Zhang

The advancements in Large Language Models (LLMs) have been hindered by their substantial sizes, which necessitate LLM compression methods for practical deployment. Singular Value Decomposition (SVD) offers a promising solution for LLM compression. However, state-of-the-art SVD-based LLM compression methods have two key limitations: truncating smaller singular values may lead to higher compression loss, and the lack of update on the compressed weight after SVD truncation. In this work, we propose SVD-LLM, a new SVD-based LLM compression method that addresses the limitations of existing methods. SVD-LLM incorporates a truncation-aware data whitening strategy to ensure a direct mapping between singular values and compression loss. Moreover, SVD-LLM adopts a layer-wise closed-form model parameter update strategy to compensate for accuracy degradation under high compression ratios. We evaluate SVD-LLM on a total of 10 datasets and eight models from three different LLM families at four different scales. Our results demonstrate the superiority of SVD-LLM over state-of-the-arts, especially at high model compression ratios.

http://arxiv.org/abs/2403.07378v3

cs.CL, cs.LG

NavCoT: Boosting LLM-Based Vision-and-Language Navigation via Learning Disentangled Reasoning

Bingqian Lin, Yunshuang Nie, Ziming Wei, Jiaqi Chen, Shikui Ma, Jianhua Han, Hang Xu, Xiaojun Chang, Xiaodan Liang

Vision-and-Language Navigation (VLN), as a crucial research problem of Embodied AI, requires an embodied agent to navigate through complex 3D environments following natural language instructions. Recent research has highlighted the promising capacity of large language models (LLMs) in VLN by improving navigational reasoning accuracy and interpretability. However, their predominant use in an offline manner usually suffers from substantial domain gap between the VLN task and the LLM training corpus. This paper introduces a novel strategy called Navigational Chain-of-Thought (NavCoT), where we fulfill parameter-efficient in-domain training to enable self-guided navigational decision, leading to a significant mitigation of the domain gap in a cost-effective manner. Specifically, at each timestep, the LLM is prompted to forecast the navigational chain-of-thought by: 1) acting as a world model to imagine the next observation according to the instruction, 2) selecting the candidate observation that best aligns with the imagination, and 3) determining the action based on the reasoning from the prior steps. Through constructing formalized labels for training, the LLM can learn to generate desired and reasonable chain-of-thought outputs for improving the action decision. Experimental results across various training settings and popular VLN benchmarks (e.g., Room-to-Room (R2R), Room-across-Room (RxR), Room-for-Room (R4R)) show the significant superiority of NavCoT over the direct action prediction variants. Through simple parameter-efficient finetuning, our NavCoT outperforms a recent GPT4-based approach with ~7% relative improvement on the R2R dataset. We believe that NavCoT will help unlock more task-adaptive and scalable LLM-based embodied agents, which are helpful for developing real-world robotics applications. Code is available at https://github.com/expectorlin/NavCoT.

http://arxiv.org/abs/2403.07376v1

cs.CV, cs.AI, cs.CL, cs.RO

CKERC : Joint Large Language Models with Commonsense Knowledge for Emotion Recognition in Conversation

Yumeng Fu

Emotion recognition in conversation (ERC) is a task which predicts the emotion of an utterance in the context of a conversation. It tightly depends on dialogue context, speaker identity information, multiparty dialogue scenario and so on. However, the state-of-the-art method (instructERC) solely identifying speaker, and ignores commonsense knowledge(i.e., reaction of the listeners and intention of the speaker, etc.) behind speakers during a conversation, which can deeply mine speaker information. To this end, we propose a novel joint large language models with commonsense knowledge framework for emotion recognition in conversation, namely CKERC.We design prompts to generate interlocutors' commonsense based on historical utterances with large language model. And we use the interlocutor commonsense identification task for LLM pre-training to fine-tune speaker implicit clues information.By solving above challenge, our method achieve state-of-the-art.We extensive experiment on three widely-used datasets, i.e., IEMOCAP, MELD, EmoryNLP, demonstrate our method superiority. Also, we conduct in-depth analysis and further demonstrate the effectiveness of commonsense knowledge in ERC task in large language model.

http://arxiv.org/abs/2403.07260v1

cs.CL

Curry-DPO: Enhancing Alignment using Curriculum Learning & Ranked Preferences

Pulkit Pattnaik, Rishabh Maheshwary, Kelechi Ogueji, Vikas Yadav, Sathwik Tejaswi Madhusudhan

Direct Preference Optimization (DPO) is an effective technique that leverages pairwise preference data (usually one chosen and rejected response pair per user prompt) to align LLMs to human preferences. In practice, multiple responses can exist for a given prompt with varying quality relative to each other. With availability of such quality ratings for multiple responses, we propose utilizing these responses to create multiple preference pairs for a given prompt. Our work focuses on systematically using the constructed multiple preference pair in DPO training via curriculum learning methodology. In particular, we order these multiple pairs of preference data from easy to hard (emulating curriculum training) according to various criteria. We show detailed comparisons of our proposed approach to the standard single-pair DPO setting. Our method, which we call Curry-DPO consistently shows increased performance gains on MTbench, Vicuna, WizardLM, and the UltraFeedback test set, highlighting its effectiveness. More specifically, Curry-DPO achieves a score of 7.43 on MT-bench with Zephy-7B model outperforming majority of existing LLMs with similar parameter size. Curry-DPO also achieves the highest adjusted win rates on Vicuna, WizardLM, and UltraFeedback test datasets (90.7%, 87.1%, and 87.9% respectively) in our experiments, with notable gains of upto 7.5% when compared to standard DPO technique.

http://arxiv.org/abs/2403.07230v1

cs.CL, cs.AI, cs.LG

Action Reimagined: Text-to-Pose Video Editing for Dynamic Human Actions

Lan Wang, Vishnu Boddeti, Sernam Lim

We introduce a novel text-to-pose video editing method, ReimaginedAct. While existing video editing tasks are limited to changes in attributes, backgrounds, and styles, our method aims to predict open-ended human action changes in video. Moreover, our method can accept not only direct instructional text prompts but also `what if' questions to predict possible action changes. ReimaginedAct comprises video understanding, reasoning, and editing modules. First, an LLM is utilized initially to obtain a plausible answer for the instruction or question, which is then used for (1) prompting Grounded-SAM to produce bounding boxes of relevant individuals and (2) retrieving a set of pose videos that we have collected for editing human actions. The retrieved pose videos and the detected individuals are then utilized to alter the poses extracted from the original video. We also employ a timestep blending module to ensure the edited video retains its original content except where necessary modifications are needed. To facilitate research in text-to-pose video editing, we introduce a new evaluation dataset, WhatifVideo-1.0. This dataset includes videos of different scenarios spanning a range of difficulty levels, along with questions and text prompts. Experimental results demonstrate that existing video editing methods struggle with human action editing, while our approach can achieve effective action editing and even imaginary editing from counterfactual questions.

http://arxiv.org/abs/2403.07198v1

cs.CV

UPS: Towards Foundation Models for PDE Solving via Cross-Modal Adaptation

Junhong Shen, Tanya Marwah, Ameet Talwalkar

We introduce UPS (Unified PDE Solver), an effective and data-efficient approach to solve diverse spatiotemporal PDEs defined over various domains, dimensions, and resolutions. UPS unifies different PDEs into a consistent representation space and processes diverse collections of PDE data using a unified network architecture that combines LLMs with domain-specific neural operators. We train the network via a two-stage cross-modal adaptation process, leveraging ideas of modality alignment and multi-task learning. By adapting from pretrained LLMs and exploiting text-form meta information, we are able to use considerably fewer training samples than previous methods while obtaining strong empirical results. UPS outperforms existing baselines, often by a large margin, on a wide range of 1D and 2D datasets in PDEBench, achieving state-of-the-art results on 8 of 10 tasks considered. Meanwhile, it is capable of few-shot transfer to different PDE families, coefficients, and resolutions.

http://arxiv.org/abs/2403.07187v1

cs.LG

SMART: Automatically Scaling Down Language Models with Accuracy Guarantees for Reduced Processing Fees

Saehan Jo, Immanuel Trummer

The advancement of Large Language Models (LLMs) has significantly boosted performance in natural language processing (NLP) tasks. However, the deployment of high-performance LLMs incurs substantial costs, primarily due to the increased number of parameters aimed at enhancing model performance. This has made the use of state-of-the-art LLMs more expensive for end-users. AI service providers, such as OpenAI and Anthropic, often offer multiple versions of LLMs with varying prices and performance. However, end-users still face challenges in choosing the appropriate LLM for their tasks that balance result quality with cost. We introduce SMART, Scaling Models Adaptively for Reduced Token Fees, a novel LLM framework designed to minimize the inference costs of NLP tasks while ensuring sufficient result quality. It enables users to specify an accuracy constraint in terms of the equivalence of outputs to those of the most powerful LLM. SMART then generates results that deviate from the outputs of this LLM only with a probability below a user-defined threshold. SMART employs a profiling phase that evaluates the performance of multiple LLMs to identify those that meet the user-defined accuracy level. SMART optimizes the tradeoff between profiling overheads and the anticipated cost savings resulting from profiling. Moreover, our approach significantly reduces inference costs by strategically leveraging a mix of LLMs. Our experiments on three real-world datasets show that, based on OpenAI models, SMART achieves significant cost savings, up to 25.6x in comparison to GPT-4.

http://arxiv.org/abs/2403.13835v1

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

SELMA: Learning and Merging Skill-Specific Text-to-Image Experts with Auto-Generated Data

Jialu Li, Jaemin Cho, Yi-Lin Sung, Jaehong Yoon, Mohit Bansal

Recent text-to-image (T2I) generation models have demonstrated impressive capabilities in creating images from text descriptions. However, these T2I generation models often fall short of generating images that precisely match the details of the text inputs, such as incorrect spatial relationship or missing objects. In this paper, we introduce SELMA: Skill-Specific Expert Learning and Merging with Auto-Generated Data, a novel paradigm to improve the faithfulness of T2I models by fine-tuning models on automatically generated, multi-skill image-text datasets, with skill-specific expert learning and merging. First, SELMA leverages an LLM's in-context learning capability to generate multiple datasets of text prompts that can teach different skills, and then generates the images with a T2I model based on the prompts. Next, SELMA adapts the T2I model to the new skills by learning multiple single-skill LoRA (low-rank adaptation) experts followed by expert merging. Our independent expert fine-tuning specializes multiple models for different skills, and expert merging helps build a joint multi-skill T2I model that can generate faithful images given diverse text prompts, while mitigating the knowledge conflict from different datasets. We empirically demonstrate that SELMA significantly improves the semantic alignment and text faithfulness of state-of-the-art T2I diffusion models on multiple benchmarks (+2.1% on TIFA and +6.9% on DSG), human preference metrics (PickScore, ImageReward, and HPS), as well as human evaluation. Moreover, fine-tuning with image-text pairs auto-collected via SELMA shows comparable performance to fine-tuning with ground truth data. Lastly, we show that fine-tuning with images from a weaker T2I model can help improve the generation quality of a stronger T2I model, suggesting promising weak-to-strong generalization in T2I models.

http://arxiv.org/abs/2403.06952v1

cs.CV, cs.AI, cs.CL, cs.LG

Naming, Describing, and Quantifying Visual Objects in Humans and LLMs

Alberto Testoni, Juell Sprott, Sandro Pezzelle

While human speakers use a variety of different expressions when describing the same object in an image, giving rise to a distribution of plausible labels driven by pragmatic constraints, the extent to which current Vision \& Language Large Language Models (VLLMs) can mimic this crucial feature of language use is an open question. This applies to common, everyday objects, but it is particularly interesting for uncommon or novel objects for which a category label may be lacking or fuzzy. Furthermore, humans show clear production preferences for highly context-sensitive expressions, such as the quantifiers `few' or `most'. In our work, we evaluate VLLMs (FROMAGe, BLIP-2, LLaVA) on three categories (nouns, attributes, and quantifiers) where humans show great subjective variability concerning the distribution over plausible labels, using datasets and resources mostly under-explored in previous work. Our results reveal mixed evidence on the ability of VLLMs to capture human naming preferences, with all models failing in tasks that require high-level reasoning such as assigning quantifiers.

http://arxiv.org/abs/2403.06935v2

cs.CL

FocusCLIP: Multimodal Subject-Level Guidance for Zero-Shot Transfer in Human-Centric Tasks

Muhammad Saif Ullah Khan, Muhammad Ferjad Naeem, Federico Tombari, Luc Van Gool, Didier Stricker, Muhammad Zeshan Afzal

We propose FocusCLIP, integrating subject-level guidance--a specialized mechanism for target-specific supervision--into the CLIP framework for improved zero-shot transfer on human-centric tasks. Our novel contributions enhance CLIP on both the vision and text sides. On the vision side, we incorporate ROI heatmaps emulating human visual attention mechanisms to emphasize subject-relevant image regions. On the text side, we introduce human pose descriptions to provide rich contextual information. For human-centric tasks, FocusCLIP is trained with images from the MPII Human Pose dataset. The proposed approach surpassed CLIP by an average of 8.61% across five previously unseen datasets covering three human-centric tasks. FocusCLIP achieved an average accuracy of 33.65% compared to 25.04% by CLIP. We observed a 3.98% improvement in activity recognition, a 14.78% improvement in age classification, and a 7.06% improvement in emotion recognition. Moreover, using our proposed single-shot LLM prompting strategy, we release a high-quality MPII Pose Descriptions dataset to encourage further research in multimodal learning for human-centric tasks. Furthermore, we also demonstrate the effectiveness of our subject-level supervision on non-human-centric tasks. FocusCLIP shows a 2.47% improvement over CLIP in zero-shot bird classification using the CUB dataset. Our findings emphasize the potential of integrating subject-level guidance with general pretraining methods for enhanced downstream performance.

http://arxiv.org/abs/2403.06904v2

cs.CV

Exploring Large Language Models and Hierarchical Frameworks for Classification of Large Unstructured Legal Documents

Nishchal Prasad, Mohand Boughanem, Taoufiq Dkaki

Legal judgment prediction suffers from the problem of long case documents exceeding tens of thousands of words, in general, and having a non-uniform structure. Predicting judgments from such documents becomes a challenging task, more so on documents with no structural annotation. We explore the classification of these large legal documents and their lack of structural information with a deep-learning-based hierarchical framework which we call MESc; "Multi-stage Encoder-based Supervised with-clustering"; for judgment prediction. Specifically, we divide a document into parts to extract their embeddings from the last four layers of a custom fine-tuned Large Language Model, and try to approximate their structure through unsupervised clustering. Which we use in another set of transformer encoder layers to learn the inter-chunk representations. We analyze the adaptability of Large Language Models (LLMs) with multi-billion parameters (GPT-Neo, and GPT-J) with the hierarchical framework of MESc and compare them with their standalone performance on legal texts. We also study their intra-domain(legal) transfer learning capability and the impact of combining embeddings from their last layers in MESc. We test these methods and their effectiveness with extensive experiments and ablation studies on legal documents from India, the European Union, and the United States with the ILDC dataset and a subset of the LexGLUE dataset. Our approach achieves a minimum total performance gain of approximately 2 points over previous state-of-the-art methods.

http://arxiv.org/abs/2403.06872v1

cs.CL, cs.AI

ACFIX: Guiding LLMs with Mined Common RBAC Practices for Context-Aware Repair of Access Control Vulnerabilities in Smart Contracts

Lyuye Zhang, Kaixuan Li, Kairan Sun, Daoyuan Wu, Ye Liu, Haoye Tian, Yang Liu

Smart contracts are susceptible to various security issues, among which access control (AC) vulnerabilities are particularly critical. While existing research has proposed multiple detection tools, the automatic and appropriate repair of AC vulnerabilities in smart contracts remains a challenge. Unlike commonly supported vulnerability types by existing repair tools, such as reentrancy, which are usually fixed by template-based approaches, the main obstacle of AC lies in identifying the appropriate roles or permissions amid a long list of non-AC-related source code to generate proper patch code, a task that demands human-level intelligence. Leveraging recent advancements in large language models (LLMs), we employ the state-of-the-art GPT-4 model and enhance it with a novel approach called ACFIX. The key insight is that we can mine common AC practices for major categories of code functionality and use them to guide LLMs in fixing code with similar functionality. To this end, ACFIX involves both offline and online phases. First, during the offline phase, ACFIX mines a taxonomy of common Role-based Access Control (RBAC) practices from 344,251 on-chain contracts, categorizing 49 role-permission pairs from the top 1,000 pairs mined. Second, during the online phase, ACFIX tracks AC-related elements across the contract and uses this context information along with a Chain-of-Thought pipeline to guide LLMs in identifying the most appropriate role-permission pair for the subject contract and subsequently generating a suitable patch. This patch will then undergo a validity and effectiveness check. To evaluate ACFIX, we built the first benchmark dataset of 118 real-world AC vulnerabilities, and our evaluation revealed that ACFIX successfully repaired 94.92% of them. This represents a significant improvement compared to the baseline GPT-4, which achieved only 52.54%.

http://arxiv.org/abs/2403.06838v2

cs.SE, cs.CR

Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?

Egor Zverev, Sahar Abdelnabi, Mario Fritz, Christoph H. Lampert

Instruction-tuned Large Language Models (LLMs) have achieved breakthrough results, opening countless new possibilities for many practical applications. However, LLMs lack elementary safety features that are established norms in other areas of computer science, such as the separation between instructions and data, causing them to malfunction or rendering them vulnerable to manipulation and interference by third parties e.g., via indirect prompt/command injection. Even worse, so far, there is not even an established definition of what precisely such a separation would mean and how its violation could be tested. In this work, we aim to close this gap. We introduce a formal measure to quantify the phenomenon of instruction-data separation as well as an empirical variant of the measure that can be computed from a model`s black-box outputs. We also introduce a new dataset, SEP (Should it be Executed or Processed?), which allows estimating the measure, and we report results on several state-of-the-art open-source and closed LLMs. Finally, we quantitatively demonstrate that all evaluated LLMs fail to achieve a high amount of separation, according to our measure. The source code and SEP dataset are openly accessible at https://github.com/egozverev/Shold-It-Be-Executed-Or-Processed.

http://arxiv.org/abs/2403.06833v1

cs.LG, cs.CL

The Power of Noise: Toward a Unified Multi-modal Knowledge Graph Representation Framework

Zhuo Chen, Yin Fang, Yichi Zhang, Lingbing Guo, Jiaoyan Chen, Huajun Chen, Wen Zhang

The advancement of Multi-modal Pre-training highlights the necessity for a robust Multi-Modal Knowledge Graph (MMKG) representation learning framework. This framework is crucial for integrating structured knowledge into multi-modal Large Language Models (LLMs) at scale, aiming to alleviate issues like knowledge misconceptions and multi-modal hallucinations. In this work, to evaluate models' ability to accurately embed entities within MMKGs, we focus on two widely researched tasks: Multi-modal Knowledge Graph Completion (MKGC) and Multi-modal Entity Alignment (MMEA). Building on this foundation, we propose a novel SNAG method that utilizes a Transformer-based architecture equipped with modality-level noise masking for the robust integration of multi-modal entity features in KGs. By incorporating specific training objectives for both MKGC and MMEA, our approach achieves SOTA performance across a total of ten datasets (three for MKGC and seven for MEMA), demonstrating its robustness and versatility. Besides, SNAG can not only function as a standalone model but also enhance other existing methods, providing stable performance improvements. Our code and data are available at: https://github.com/zjukg/SNAG.

http://arxiv.org/abs/2403.06832v2

cs.CL, cs.AI

ConspEmoLLM: Conspiracy Theory Detection Using an Emotion-Based Large Language Model

Zhiwei Liu, Boyang Liu, Paul Thompson, Kailai Yang, Raghav Jain, Sophia Ananiadou

The internet has brought both benefits and harms to society. A prime example of the latter is misinformation, including conspiracy theories, which flood the web. Recent advances in natural language processing, particularly the emergence of large language models (LLMs), have improved the prospects of accurate misinformation detection. However, most LLM-based approaches to conspiracy theory detection focus only on binary classification and fail to account for the important relationship between misinformation and affective features (i.e., sentiment and emotions). Driven by a comprehensive analysis of conspiracy text that reveals its distinctive affective features, we propose ConspEmoLLM, the first open-source LLM that integrates affective information and is able to perform diverse tasks relating to conspiracy theories. These tasks include not only conspiracy theory detection, but also classification of theory type and detection of related discussion (e.g., opinions towards theories). ConspEmoLLM is fine-tuned based on an emotion-oriented LLM using our novel ConDID dataset, which includes five tasks to support LLM instruction tuning and evaluation. We demonstrate that when applied to these tasks, ConspEmoLLM largely outperforms several open-source general domain LLMs and ChatGPT, as well as an LLM that has been fine-tuned using ConDID, but which does not use affective features. This project will be released on https://github.com/lzw108/ConspEmoLLM/.

http://arxiv.org/abs/2403.06765v1

cs.CL

HILL: A Hallucination Identifier for Large Language Models

Florian Leiser, Sven Eckhardt, Valentin Leuthe, Merlin Knaeble, Alexander Maedche, Gerhard Schwabe, Ali Sunyaev

Large language models (LLMs) are prone to hallucinations, i.e., nonsensical, unfaithful, and undesirable text. Users tend to overrely on LLMs and corresponding hallucinations which can lead to misinterpretations and errors. To tackle the problem of overreliance, we propose HILL, the "Hallucination Identifier for Large Language Models". First, we identified design features for HILL with a Wizard of Oz approach with nine participants. Subsequently, we implemented HILL based on the identified design features and evaluated HILL's interface design by surveying 17 participants. Further, we investigated HILL's functionality to identify hallucinations based on an existing question-answering dataset and five user interviews. We find that HILL can correctly identify and highlight hallucinations in LLM responses which enables users to handle LLM responses with more caution. With that, we propose an easy-to-implement adaptation to existing LLMs and demonstrate the relevance of user-centered designs of AI artifacts.

http://arxiv.org/abs/2403.06710v1

cs.HC

Zero-Shot ECG Classification with Multimodal Learning and Test-time Clinical Knowledge Enhancement

Che Liu, Zhongwei Wan, Cheng Ouyang, Anand Shah, Wenjia Bai, Rossella Arcucci

Electrocardiograms (ECGs) are non-invasive diagnostic tools crucial for detecting cardiac arrhythmic diseases in clinical practice. While ECG Self-supervised Learning (eSSL) methods show promise in representation learning from unannotated ECG data, they often overlook the clinical knowledge that can be found in reports. This oversight and the requirement for annotated samples for downstream tasks limit eSSL's versatility. In this work, we address these issues with the Multimodal ECG Representation Learning (MERL}) framework. Through multimodal learning on ECG records and associated reports, MERL is capable of performing zero-shot ECG classification with text prompts, eliminating the need for training data in downstream tasks. At test time, we propose the Clinical Knowledge Enhanced Prompt Engineering (CKEPE) approach, which uses Large Language Models (LLMs) to exploit external expert-verified clinical knowledge databases, generating more descriptive prompts and reducing hallucinations in LLM-generated content to boost zero-shot classification. Based on MERL, we perform the first benchmark across six public ECG datasets, showing the superior performance of MERL compared against eSSL methods. Notably, MERL achieves an average AUC score of 75.2% in zero-shot classification (without training data), 3.2% higher than linear probed eSSL methods with 10\% annotated training data, averaged across all six datasets.

http://arxiv.org/abs/2403.06659v2

eess.SP, cs.AI, cs.LG

Elephants Never Forget: Testing Language Models for Memorization of Tabular Data

Sebastian Bordt, Harsha Nori, Rich Caruana

While many have shown how Large Language Models (LLMs) can be applied to a diverse set of tasks, the critical issues of data contamination and memorization are often glossed over. In this work, we address this concern for tabular data. Starting with simple qualitative tests for whether an LLM knows the names and values of features, we introduce a variety of different techniques to assess the degrees of contamination, including statistical tests for conditional distribution modeling and four tests that identify memorization. Our investigation reveals that LLMs are pre-trained on many popular tabular datasets. This exposure can lead to invalid performance evaluation on downstream tasks because the LLMs have, in effect, been fit to the test set. Interestingly, we also identify a regime where the language model reproduces important statistics of the data, but fails to reproduce the dataset verbatim. On these datasets, although seen during training, good performance on downstream tasks might not be due to overfitting. Our findings underscore the need for ensuring data integrity in machine learning tasks with LLMs. To facilitate future research, we release an open-source tool that can perform various tests for memorization \url{https://github.com/interpretml/LLM-Tabular-Memorization-Checker}.

http://arxiv.org/abs/2403.06644v1

cs.LG, cs.CL

KELLMRec: Knowledge-Enhanced Large Language Models for Recommendation

Weiqing Luo, Chonggang Song, Lingling Yi, Gong Cheng

The utilization of semantic information is an important research problem in the field of recommender systems, which aims to complement the missing parts of mainstream ID-based approaches. With the rise of LLM, its ability to act as a knowledge base and its reasoning capability have opened up new possibilities for this research area, making LLM-based recommendation an emerging research direction. However, directly using LLM to process semantic information for recommendation scenarios is unreliable and sub-optimal due to several problems such as hallucination. A promising way to cope with this is to use external knowledge to aid LLM in generating truthful and usable text. Inspired by the above motivation, we propose a Knowledge-Enhanced LLMRec method. In addition to using external knowledge in prompts, the proposed method also includes a knowledge-based contrastive learning scheme for training. Experiments on public datasets and in-enterprise datasets validate the effectiveness of the proposed method.

http://arxiv.org/abs/2403.06642v1

cs.IR, cs.AI, cs.CL

MedKP: Medical Dialogue with Knowledge Enhancement and Clinical Pathway Encoding

Jiageng Wu, Xian Wu, Yefeng Zheng, Jie Yang

With appropriate data selection and training techniques, Large Language Models (LLMs) have demonstrated exceptional success in various medical examinations and multiple-choice questions. However, the application of LLMs in medical dialogue generation-a task more closely aligned with actual medical practice-has been less explored. This gap is attributed to the insufficient medical knowledge of LLMs, which leads to inaccuracies and hallucinated information in the generated medical responses. In this work, we introduce the Medical dialogue with Knowledge enhancement and clinical Pathway encoding (MedKP) framework, which integrates an external knowledge enhancement module through a medical knowledge graph and an internal clinical pathway encoding via medical entities and physician actions. Evaluated with comprehensive metrics, our experiments on two large-scale, real-world online medical consultation datasets (MedDG and KaMed) demonstrate that MedKP surpasses multiple baselines and mitigates the incidence of hallucinations, achieving a new state-of-the-art. Extensive ablation studies further reveal the effectiveness of each component of MedKP. This enhancement advances the development of reliable, automated medical consultation responses using LLMs, thereby broadening the potential accessibility of precise and real-time medical assistance.

http://arxiv.org/abs/2403.06611v1

cs.CL, cs.AI

Guiding Clinical Reasoning with Large Language Models via Knowledge Seeds

Jiageng WU, Xian Wu, Jie Yang

Clinical reasoning refers to the cognitive process that physicians employ in evaluating and managing patients. This process typically involves suggesting necessary examinations, diagnosing patients' diseases, and deciding on appropriate therapies, etc. Accurate clinical reasoning requires extensive medical knowledge and rich clinical experience, setting a high bar for physicians. This is particularly challenging in developing countries due to the overwhelming number of patients and limited physician resources, contributing significantly to global health inequity and necessitating automated clinical reasoning approaches. Recently, the emergence of large language models (LLMs) such as ChatGPT and GPT-4 have demonstrated their potential in clinical reasoning. However, these LLMs are prone to hallucination problems, and the reasoning process of LLMs may not align with the clinical decision path of physicians. In this study, we introduce a novel framework, In-Context Padding (ICP), designed to enhance LLMs with medical knowledge. Specifically, we infer critical clinical reasoning elements (referred to as knowledge seeds) and use these as anchors to guide the generation process of LLMs. Experiments on two clinical question datasets demonstrate that ICP significantly improves the clinical reasoning ability of LLMs.

http://arxiv.org/abs/2403.06609v1

cs.CL, cs.AI

Decoding Complexity: Exploring Human-AI Concordance in Qualitative Coding

Elisabeth Kirsten, Annalina Buckmann, Abraham Mhaidli, Steffen Becker

Qualitative data analysis provides insight into the underlying perceptions and experiences within unstructured data. However, the time-consuming nature of the coding process, especially for larger datasets, calls for innovative approaches, such as the integration of Large Language Models (LLMs). This short paper presents initial findings from a study investigating the integration of LLMs for coding tasks of varying complexity in a real-world dataset. Our results highlight the challenges inherent in coding with extensive codebooks and contexts, both for human coders and LLMs, and suggest that the integration of LLMs into the coding process requires a task-by-task evaluation. We examine factors influencing the complexity of coding tasks and initiate a discussion on the usefulness and limitations of incorporating LLMs in qualitative research.

http://arxiv.org/abs/2403.06607v1

cs.HC

ContextGPT: Infusing LLMs Knowledge into Neuro-Symbolic Activity Recognition Models

Luca Arrotta, Claudio Bettini, Gabriele Civitarese, Michele Fiori

Context-aware Human Activity Recognition (HAR) is a hot research area in mobile computing, and the most effective solutions in the literature are based on supervised deep learning models. However, the actual deployment of these systems is limited by the scarcity of labeled data that is required for training. Neuro-Symbolic AI (NeSy) provides an interesting research direction to mitigate this issue, by infusing common-sense knowledge about human activities and the contexts in which they can be performed into HAR deep learning classifiers. Existing NeSy methods for context-aware HAR rely on knowledge encoded in logic-based models (e.g., ontologies) whose design, implementation, and maintenance to capture new activities and contexts require significant human engineering efforts, technical knowledge, and domain expertise. Recent works show that pre-trained Large Language Models (LLMs) effectively encode common-sense knowledge about human activities. In this work, we propose ContextGPT: a novel prompt engineering approach to retrieve from LLMs common-sense knowledge about the relationship between human activities and the context in which they are performed. Unlike ontologies, ContextGPT requires limited human effort and expertise. An extensive evaluation carried out on two public datasets shows how a NeSy model obtained by infusing common-sense knowledge from ContextGPT is effective in data scarcity scenarios, leading to similar (and sometimes better) recognition rates than logic-based approaches with a fraction of the effort.

http://arxiv.org/abs/2403.06586v1

cs.LG, cs.AI, cs.CL

From English to ASIC: Hardware Implementation with Large Language Model

Emil Goh, Maoyang Xiang, I-Chyn Wey, T. Hui Teo

In the realm of ASIC engineering, the landscape has been significantly reshaped by the rapid development of LLM, paralleled by an increase in the complexity of modern digital circuits. This complexity has escalated the requirements for HDL coding, necessitating a higher degree of precision and sophistication. However, challenges have been faced due to the less-than-optimal performance of modern language models in generating hardware description code, a situation further exacerbated by the scarcity of the corresponding high-quality code datasets. These challenges have highlighted the gap between the potential of LLMs to revolutionize digital circuit design and their current capabilities in accurately interpreting and implementing hardware specifications. To address these challenges, a strategy focusing on the fine-tuning of the leading-edge nature language model and the reshuffling of the HDL code dataset has been developed. The fine-tuning aims to enhance models' proficiency in generating precise and efficient ASIC design, while the dataset reshuffling is intended to broaden the scope and improve the quality of training material. The model demonstrated significant improvements compared to the base model, with approximately 10% to 20% increase in accuracy across a wide range of temperature for the pass@1 metric. This approach is expected to facilitate a simplified and more efficient LLM-assisted framework for complex circuit design, leveraging their capabilities to meet the sophisticated demands of HDL coding and thus streamlining the ASIC development process.

http://arxiv.org/abs/2403.07039v1

cs.AR, cs.AI, cs.CL

On the Consideration of AI Openness: Can Good Intent Be Abused?

Yeeun Kim, Eunkyung Choi, Hyunjun Kim, Hongseok Oh, Hyunseo Shin, Wonseok Hwang

Openness is critical for the advancement of science. In particular, recent rapid progress in AI has been made possible only by various open-source models, datasets, and libraries. However, this openness also means that technologies can be freely used for socially harmful purposes. Can open-source models or datasets be used for malicious purposes? If so, how easy is it to adapt technology for such goals? Here, we conduct a case study in the legal domain, a realm where individual decisions can have profound social consequences. To this end, we build EVE, a dataset consisting of 200 examples of questions and corresponding answers about criminal activities based on 200 Korean precedents. We found that a widely accepted open-source LLM, which initially refuses to answer unethical questions, can be easily tuned with EVE to provide unethical and informative answers about criminal activities. This implies that although open-source technologies contribute to scientific progress, some care must be taken to mitigate possible malicious use cases. Warning: This paper contains contents that some may find unethical.

http://arxiv.org/abs/2403.06537v1

cs.CL

Knowledge-aware Alert Aggregation in Large-scale Cloud Systems: a Hybrid Approach

Jinxi Kuang, Jinyang Liu, Junjie Huang, Renyi Zhong, Jiazhen Gu, Lan Yu, Rui Tan, Zengyin Yang, Michael R. Lyu

Due to the scale and complexity of cloud systems, a system failure would trigger an "alert storm", i.e., massive correlated alerts. Although these alerts can be traced back to a few root causes, the overwhelming number makes it infeasible for manual handling. Alert aggregation is thus critical to help engineers concentrate on the root cause and facilitate failure resolution. Existing methods typically utilize semantic similarity-based methods or statistical methods to aggregate alerts. However, semantic similarity-based methods overlook the causal rationale of alerts, while statistical methods can hardly handle infrequent alerts. To tackle these limitations, we introduce leveraging external knowledge, i.e., Standard Operation Procedure (SOP) of alerts as a supplement. We propose COLA, a novel hybrid approach based on correlation mining and LLM (Large Language Model) reasoning for online alert aggregation. The correlation mining module effectively captures the temporal and spatial relations between alerts, measuring their correlations in an efficient manner. Subsequently, only uncertain pairs with low confidence are forwarded to the LLM reasoning module for detailed analysis. This hybrid design harnesses both statistical evidence for frequent alerts and the reasoning capabilities of computationally intensive LLMs, ensuring the overall efficiency of COLA in handling large volumes of alerts in practical scenarios. We evaluate COLA on three datasets collected from the production environment of a large-scale cloud platform. The experimental results show COLA achieves F1-scores from 0.901 to 0.930, outperforming state-of-the-art methods and achieving comparable efficiency. We also share our experience in deploying COLA in our real-world cloud system, Cloud X.

http://arxiv.org/abs/2403.06485v1

cs.SE, cs.CL, cs.LG

CoRAL: Collaborative Retrieval-Augmented Large Language Models Improve Long-tail Recommendation

Junda Wu, Cheng-Chun Chang, Tong Yu, Zhankui He, Jianing Wang, Yupeng Hou, Julian McAuley

The long-tail recommendation is a challenging task for traditional recommender systems, due to data sparsity and data imbalance issues. The recent development of large language models (LLMs) has shown their abilities in complex reasoning, which can help to deduce users' preferences based on very few previous interactions. However, since most LLM-based systems rely on items' semantic meaning as the sole evidence for reasoning, the collaborative information of user-item interactions is neglected, which can cause the LLM's reasoning to be misaligned with task-specific collaborative information of the dataset. To further align LLMs' reasoning to task-specific user-item interaction knowledge, we introduce collaborative retrieval-augmented LLMs, CoRAL, which directly incorporate collaborative evidence into the prompts. Based on the retrieved user-item interactions, the LLM can analyze shared and distinct preferences among users, and summarize the patterns indicating which types of users would be attracted by certain items. The retrieved collaborative evidence prompts the LLM to align its reasoning with the user-item interaction patterns in the dataset. However, since the capacity of the input prompt is limited, finding the minimally-sufficient collaborative information for recommendation tasks can be challenging. We propose to find the optimal interaction set through a sequential decision-making process and develop a retrieval policy learned through a reinforcement learning (RL) framework, CoRAL. Our experimental results show that CoRAL can significantly improve LLMs' reasoning abilities on specific recommendation tasks. Our analysis also reveals that CoRAL can more efficiently explore collaborative information through reinforcement learning.

http://arxiv.org/abs/2403.06447v1

cs.IR, cs.AI

CLIcK: A Benchmark Dataset of Cultural and Linguistic Intelligence in Korean

Eunsu Kim, Juyoung Suk, Philhoon Oh, Haneul Yoo, James Thorne, Alice Oh

Despite the rapid development of large language models (LLMs) for the Korean language, there remains an obvious lack of benchmark datasets that test the requisite Korean cultural and linguistic knowledge. Because many existing Korean benchmark datasets are derived from the English counterparts through translation, they often overlook the different cultural contexts. For the few benchmark datasets that are sourced from Korean data capturing cultural knowledge, only narrow tasks such as bias and hate speech detection are offered. To address this gap, we introduce a benchmark of Cultural and Linguistic Intelligence in Korean (CLIcK), a dataset comprising 1,995 QA pairs. CLIcK sources its data from official Korean exams and textbooks, partitioning the questions into eleven categories under the two main categories of language and culture. For each instance in CLIcK, we provide fine-grained annotation of which cultural and linguistic knowledge is required to answer the question correctly. Using CLIcK, we test 13 language models to assess their performance. Our evaluation uncovers insights into their performances across the categories, as well as the diverse factors affecting their comprehension. CLIcK offers the first large-scale comprehensive Korean-centric analysis of LLMs' proficiency in Korean culture and language.

http://arxiv.org/abs/2403.06412v3

cs.CL

Can LLMs' Tuning Methods Work in Medical Multimodal Domain?

Jiawei Chen, Yue Jiang, Dingkang Yang, Mingcheng Li, Jinjie Wei, Ziyun Qian, Lihua Zhang

While large language models (LLMs) excel in world knowledge understanding, adapting them to specific subfields requires precise adjustments. Due to the model's vast scale, traditional global fine-tuning methods for large models can be computationally expensive and impact generalization. To address this challenge, a range of innovative Parameters-Efficient Fine-Tuning (PEFT) methods have emerged and achieved remarkable success in both LLMs and Large Vision-Language Models (LVLMs). In the medical domain, fine-tuning a medical Vision-Language Pretrained (VLP) model is essential for adapting it to specific tasks. Can the fine-tuning methods for large models be transferred to the medical field to enhance transfer learning efficiency? In this paper, we delve into the fine-tuning methods of LLMs and conduct extensive experiments to investigate the impact of fine-tuning methods for large models on existing multimodal models in the medical domain from the training data level and the model structure level. We show the different impacts of fine-tuning methods for large models on medical VLMs and develop the most efficient ways to fine-tune medical VLP models. We hope this research can guide medical domain researchers in optimizing VLMs' training costs, fostering the broader application of VLMs in healthcare fields. Code and dataset will be released upon acceptance.

http://arxiv.org/abs/2403.06407v1

cs.CV

Amharic LLaMA and LLaVA: Multimodal LLMs for Low Resource Languages

Michael Andersland

Large Language Models (LLMs) like GPT-4 and LLaMA have shown incredible proficiency at natural language processing tasks and have even begun to excel at tasks across other modalities such as vision and audio. Despite their success, LLMs often struggle to perform well on low-resource languages because there is so little training data available. This shortcoming is especially prevalent with open source models. In this work, we explore training LLaMA-2 to speak Amharic, a language which is spoken by over 50 million people world wide, but has orders of magnitude less data available than languages like English. We employ methods previously used for training LLMs on other languages with data scarcity, and use open source translation models to perform data augmentation and grow our dataset from millions of tokens to billions. We further enhance the capabilities of our model by connecting an image encoder and training on a translated visual instruction tuning dataset in the same manner as LLaVA, resulting in a multimodal Amharic LLM that can understand images along with text. We introduce an Amharic version of a popular benchmarking dataset to evaluate our work. Our models and dataset are open sourced and available on GitHub.

http://arxiv.org/abs/2403.06354v1

cs.CL

IndicLLMSuite: A Blueprint for Creating Pre-training and Fine-Tuning Datasets for Indian Languages

Mohammed Safi Ur Rahman Khan, Priyam Mehta, Ananth Sankar, Umashankar Kumaravelan, Sumanth Doddapaneni, Suriyaprasaad G, Varun Balan G, Sparsh Jain, Anoop Kunchukuttan, Pratyush Kumar, Raj Dabre, Mitesh M. Khapra

Despite the considerable advancements in English LLMs, the progress in building comparable models for other languages has been hindered due to the scarcity of tailored resources. Our work aims to bridge this divide by introducing an expansive suite of resources specifically designed for the development of Indic LLMs, covering 22 languages, containing a total of 251B tokens and 74.8M instruction-response pairs. Recognizing the importance of both data quality and quantity, our approach combines highly curated manually verified data, unverified yet valuable data, and synthetic data. We build a clean, open-source pipeline for curating pre-training data from diverse sources, including websites, PDFs, and videos, incorporating best practices for crawling, cleaning, flagging, and deduplication. For instruction-fine tuning, we amalgamate existing Indic datasets, translate/transliterate English datasets into Indian languages, and utilize LLaMa2 and Mixtral models to create conversations grounded in articles from Indian Wikipedia and Wikihow. Additionally, we address toxicity alignment by generating toxic prompts for multiple scenarios and then generate non-toxic responses by feeding these toxic prompts to an aligned LLaMa2 model. We hope that the datasets, tools, and resources released as a part of this work will not only propel the research and development of Indic LLMs but also establish an open-source blueprint for extending such efforts to other languages. The data and other artifacts created as part of this work are released with permissive licenses.

http://arxiv.org/abs/2403.06350v1

cs.CL

Editing Conceptual Knowledge for Large Language Models

Xiaohan Wang, Shengyu Mao, Ningyu Zhang, Shumin Deng, Yunzhi Yao, Yue Shen, Lei Liang, Jinjie Gu, Huajun Chen

Recently, there has been a growing interest in knowledge editing for Large Language Models (LLMs). Current approaches and evaluations merely explore the instance-level editing, while whether LLMs possess the capability to modify concepts remains unclear. This paper pioneers the investigation of editing conceptual knowledge for LLMs, by constructing a novel benchmark dataset ConceptEdit and establishing a suite of new metrics for evaluation. The experimental results reveal that, although existing editing methods can efficiently modify concept-level definition to some extent, they also have the potential to distort the related instantial knowledge in LLMs, leading to poor performance. We anticipate this can inspire further progress in better understanding LLMs. Our project homepage is available at https://zjunlp.github.io/project/ConceptEdit.

http://arxiv.org/abs/2403.06259v1

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

No Language is an Island: Unifying Chinese and English in Financial Large Language Models, Instruction Data, and Benchmarks

Gang Hu, Ke Qin, Chenhan Yuan, Min Peng, Alejandro Lopez-Lira, Benyou Wang, Sophia Ananiadou, Wanlong Yu, Jimin Huang, Qianqian Xie

While the progression of Large Language Models (LLMs) has notably propelled financial analysis, their application has largely been confined to singular language realms, leaving untapped the potential of bilingual Chinese-English capacity. To bridge this chasm, we introduce ICE-PIXIU, seamlessly amalgamating the ICE-INTENT model and ICE-FLARE benchmark for bilingual financial analysis. ICE-PIXIU uniquely integrates a spectrum of Chinese tasks, alongside translated and original English datasets, enriching the breadth and depth of bilingual financial modeling. It provides unrestricted access to diverse model variants, a substantial compilation of diverse cross-lingual and multi-modal instruction data, and an evaluation benchmark with expert annotations, comprising 10 NLP tasks, 20 bilingual specific tasks, totaling 95k datasets. Our thorough evaluation emphasizes the advantages of incorporating these bilingual datasets, especially in translation tasks and utilizing original English data, enhancing both linguistic flexibility and analytical acuity in financial contexts. Notably, ICE-INTENT distinguishes itself by showcasing significant enhancements over conventional LLMs and existing financial LLMs in bilingual milieus, underscoring the profound impact of robust bilingual data on the accuracy and efficacy of financial NLP.

http://arxiv.org/abs/2403.06249v2

cs.CE, cs.CL

Are You Being Tracked? Discover the Power of Zero-Shot Trajectory Tracing with LLMs!

Huanqi Yang, Sijie Ji, Rucheng Wu, Weitao Xu

There is a burgeoning discussion around the capabilities of Large Language Models (LLMs) in acting as fundamental components that can be seamlessly incorporated into Artificial Intelligence of Things (AIoT) to interpret complex trajectories. This study introduces LLMTrack, a model that illustrates how LLMs can be leveraged for Zero-Shot Trajectory Recognition by employing a novel single-prompt technique that combines role-play and think step-by-step methodologies with unprocessed Inertial Measurement Unit (IMU) data. We evaluate the model using real-world datasets designed to challenge it with distinct trajectories characterized by indoor and outdoor scenarios. In both test scenarios, LLMTrack not only meets but exceeds the performance benchmarks set by traditional machine learning approaches and even contemporary state-of-the-art deep learning models, all without the requirement of training on specialized datasets. The results of our research suggest that, with strategically designed prompts, LLMs can tap into their extensive knowledge base and are well-equipped to analyze raw sensor data with remarkable effectiveness.

http://arxiv.org/abs/2403.06201v1

cs.CL, cs.AI, cs.HC, cs.LG

Can Large Language Models Automatically Score Proficiency of Written Essays?

Watheq Mansour, Salam Albatarni, Sohaila Eltanbouly, Tamer Elsayed

Although several methods were proposed to address the problem of automated essay scoring (AES) in the last 50 years, there is still much to desire in terms of effectiveness. Large Language Models (LLMs) are transformer-based models that demonstrate extraordinary capabilities on various tasks. In this paper, we test the ability of LLMs, given their powerful linguistic knowledge, to analyze and effectively score written essays. We experimented with two popular LLMs, namely ChatGPT and Llama. We aim to check if these models can do this task and, if so, how their performance is positioned among the state-of-the-art (SOTA) models across two levels, holistically and per individual writing trait. We utilized prompt-engineering tactics in designing four different prompts to bring their maximum potential to this task. Our experiments conducted on the ASAP dataset revealed several interesting observations. First, choosing the right prompt depends highly on the model and nature of the task. Second, the two LLMs exhibited comparable average performance in AES, with a slight advantage for ChatGPT. Finally, despite the performance gap between the two LLMs and SOTA models in terms of predictions, they provide feedback to enhance the quality of the essays, which can potentially help both teachers and students.

http://arxiv.org/abs/2403.06149v2

cs.CL, cs.AI

Fine-grainedly Synthesize Streaming Data Based On Large Language Models With Graph Structure Understanding For Data Sparsity

Xin Zhang, Linhai Zhang, Deyu Zhou, Guoqiang Xu

Due to the sparsity of user data, sentiment analysis on user reviews in e-commerce platforms often suffers from poor performance, especially when faced with extremely sparse user data or long-tail labels. Recently, the emergence of LLMs has introduced new solutions to such problems by leveraging graph structures to generate supplementary user profiles. However, previous approaches have not fully utilized the graph understanding capabilities of LLMs and have struggled to adapt to complex streaming data environments. In this work, we propose a fine-grained streaming data synthesis framework that categorizes sparse users into three categories: Mid-tail, Long-tail, and Extreme. Specifically, we design LLMs to comprehensively understand three key graph elements in streaming data, including Local-global Graph Understanding, Second-Order Relationship Extraction, and Product Attribute Understanding, which enables the generation of high-quality synthetic data to effectively address sparsity across different categories. Experimental results on three real datasets demonstrate significant performance improvements, with synthesized data contributing to MSE reductions of 45.85%, 3.16%, and 62.21%, respectively.

http://arxiv.org/abs/2403.06139v1

cs.CL, cs.AI

Low-dose CT Denoising with Language-engaged Dual-space Alignment

Zhihao Chen, Tao Chen, Chenhui Wang, Chuang Niu, Ge Wang, Hongming Shan

While various deep learning methods were proposed for low-dose computed tomography (CT) denoising, they often suffer from over-smoothing, blurring, and lack of explainability. To alleviate these issues, we propose a plug-and-play Language-Engaged Dual-space Alignment loss (LEDA) to optimize low-dose CT denoising models. Our idea is to leverage large language models (LLMs) to align denoised CT and normal dose CT images in both the continuous perceptual space and discrete semantic space, which is the first LLM-based scheme for low-dose CT denoising. LEDA involves two steps: the first is to pretrain an LLM-guided CT autoencoder, which can encode a CT image into continuous high-level features and quantize them into a token space to produce semantic tokens derived from the LLM's vocabulary; and the second is to minimize the discrepancy between the denoised CT images and normal dose CT in terms of both encoded high-level features and quantized token embeddings derived by the LLM-guided CT autoencoder. Extensive experimental results on two public LDCT denoising datasets demonstrate that our LEDA can enhance existing denoising models in terms of quantitative metrics and qualitative evaluation, and also provide explainability through language-level image understanding. Source code is available at https://github.com/hao1635/LEDA.

http://arxiv.org/abs/2403.06128v1

eess.IV, cs.CV

Can LLM Substitute Human Labeling? A Case Study of Fine-grained Chinese Address Entity Recognition Dataset for UAV Delivery

Yuxuan Yao, Sichun Luo, Haohan Zhao, Guanzhi Deng, Linqi Song

We present CNER-UAV, a fine-grained \textbf{C}hinese \textbf{N}ame \textbf{E}ntity \textbf{R}ecognition dataset specifically designed for the task of address resolution in \textbf{U}nmanned \textbf{A}erial \textbf{V}ehicle delivery systems. The dataset encompasses a diverse range of five categories, enabling comprehensive training and evaluation of NER models. To construct this dataset, we sourced the data from a real-world UAV delivery system and conducted a rigorous data cleaning and desensitization process to ensure privacy and data integrity. The resulting dataset, consisting of around 12,000 annotated samples, underwent human experts and \textbf{L}arge \textbf{L}anguage \textbf{M}odel annotation. We evaluated classical NER models on our dataset and provided in-depth analysis. The dataset and models are publicly available at \url{https://github.com/zhhvvv/CNER-UAV}.

http://arxiv.org/abs/2403.06097v2

cs.CL, cs.AI, cs.IR

KG-Rank: Enhancing Large Language Models for Medical QA with Knowledge Graphs and Ranking Techniques

Rui Yang, Haoran Liu, Edison Marrese-Taylor, Qingcheng Zeng, Yu He Ke, Wanxin Li, Lechao Cheng, Qingyu Chen, James Caverlee, Yutaka Matsuo, Irene Li

Large Language Models (LLMs) have significantly advanced healthcare innovation on generation capabilities. However, their application in real clinical settings is challenging due to potential deviations from medical facts and inherent biases. In this work, we develop an augmented LLM framework, KG-Rank, which leverages a medical knowledge graph (KG) with ranking and re-ranking techniques, aiming to improve free-text question-answering (QA) in the medical domain. Specifically, upon receiving a question, we initially retrieve triplets from a medical KG to gather factual information. Subsequently, we innovatively apply ranking methods to refine the ordering of these triplets, aiming to yield more precise answers. To the best of our knowledge, KG-Rank is the first application of ranking models combined with KG in medical QA specifically for generating long answers. Evaluation of four selected medical QA datasets shows that KG-Rank achieves an improvement of over 18% in the ROUGE-L score. Moreover, we extend KG-Rank to open domains, where it realizes a 14% improvement in ROUGE-L, showing the effectiveness and potential of KG-Rank.

http://arxiv.org/abs/2403.05881v2

cs.CL

Reverse That Number! Decoding Order Matters in Arithmetic Learning

Daniel Zhang-Li, Nianyi Lin, Jifan Yu, Zheyuan Zhang, Zijun Yao, Xiaokang Zhang, Lei Hou, Jing Zhang, Juanzi Li

Recent advancements in pretraining have demonstrated that modern Large Language Models (LLMs) possess the capability to effectively learn arithmetic operations. However, despite acknowledging the significance of digit order in arithmetic computation, current methodologies predominantly rely on sequential, step-by-step approaches for teaching LLMs arithmetic, resulting in a conclusion where obtaining better performance involves fine-grained step-by-step. Diverging from this conventional path, our work introduces a novel strategy that not only reevaluates the digit order by prioritizing output from the least significant digit but also incorporates a step-by-step methodology to substantially reduce complexity. We have developed and applied this method in a comprehensive set of experiments. Compared to the previous state-of-the-art (SOTA) method, our findings reveal an overall improvement of in accuracy while requiring only a third of the tokens typically used during training. For the purpose of facilitating replication and further research, we have made our code and dataset publicly available at \url{https://anonymous.4open.science/r/RAIT-9FB7/}.

http://arxiv.org/abs/2403.05845v1

cs.CL, cs.AI

Cached Model-as-a-Resource: Provisioning Large Language Model Agents for Edge Intelligence in Space-air-ground Integrated Networks

Minrui Xu, Dusit Niyato, Hongliang Zhang, Jiawen Kang, Zehui Xiong, Shiwen Mao, Zhu Han

Space-air-ground integrated networks (SAGINs) enable worldwide network coverage beyond geographical limitations for users to access ubiquitous intelligence services. {\color{black}Facing global coverage and complex environments in SAGINs, edge intelligence can provision AI agents based on large language models (LLMs) for users via edge servers at ground base stations (BSs) or cloud data centers relayed by satellites.} As LLMs with billions of parameters are pre-trained on vast datasets, LLM agents have few-shot learning capabilities, e.g., chain-of-thought (CoT) prompting for complex tasks, which are challenged by limited resources in SAGINs. In this paper, we propose a joint caching and inference framework for edge intelligence to provision sustainable and ubiquitous LLM agents in SAGINs. We introduce "cached model-as-a-resource" for offering LLMs with limited context windows and propose a novel optimization framework, i.e., joint model caching and inference, to utilize cached model resources for provisioning LLM agent services along with communication, computing, and storage resources. We design "age of thought" (AoT) considering the CoT prompting of LLMs, and propose the least AoT cached model replacement algorithm for optimizing the provisioning cost. We propose a deep Q-network-based modified second-bid (DQMSB) auction to incentivize these network operators, which can enhance allocation efficiency while guaranteeing strategy-proofness and free from adverse selection.

http://arxiv.org/abs/2403.05826v1

cs.NI, eess.SP

Optimizing LLM Queries in Relational Workloads

Shu Liu, Asim Biswal, Audrey Cheng, Xiangxi Mo, Shiyi Cao, Joseph E. Gonzalez, Ion Stoica, Matei Zaharia

Analytical database providers (e.g., Redshift, Databricks, BigQuery) have rapidly added support for invoking Large Language Models (LLMs) through native user-defined functions (UDFs) to help users perform natural language tasks, such as classification, entity extraction, and translation, inside analytical workloads. For instance, an analyst might want to extract customer sentiments on millions of product reviews. However, LLM inference is highly expensive in both computational and economic terms: for example, an NVIDIA L4 GPU running Llama2-7B can only process 6 KB of text per second. In this paper, we explore how to optimize LLM inference for analytical workloads that invoke LLMs within relational queries. We show that relational queries present novel opportunities for accelerating LLM inference, including reordering rows to maximize key-value (KV) cache reuse within the LLM inference engine, reordering columns within a row to further increase cache reuse, and deduplicating redundant inference requests. We implement these optimizations in Apache Spark, with vLLM as the model serving backend and achieve up to 4.4x improvement in end-to-end latency on a benchmark of diverse LLM-based queries on real datasets. To the best of our knowledge, this is the first work to explicitly address the problem of optimizing LLM invocations within SQL queries.

http://arxiv.org/abs/2403.05821v1

cs.LG, cs.DB

MP2D: An Automated Topic Shift Dialogue Generation Framework Leveraging Knowledge Graphs

Yerin Hwang, Yongil Kim, Yunah Jang, Jeesoo Bang, Hyunkyung Bae, Kyomin Jung

Despite advancements in on-topic dialogue systems, effectively managing topic shifts within dialogues remains a persistent challenge, largely attributed to the limited availability of training datasets. To address this issue, we propose Multi-Passage to Dialogue (MP2D), a data generation framework that automatically creates conversational question-answering datasets with natural topic transitions. By leveraging the relationships between entities in a knowledge graph, MP2D maps the flow of topics within a dialogue, effectively mirroring the dynamics of human conversation. It retrieves relevant passages corresponding to the topics and transforms them into dialogues through the passage-to-dialogue method. Through quantitative and qualitative experiments, we demonstrate MP2D's efficacy in generating dialogue with natural topic shifts. Furthermore, this study introduces a novel benchmark for topic shift dialogues, TS-WikiDialog. Utilizing the dataset, we demonstrate that even Large Language Models (LLMs) struggle to handle topic shifts in dialogue effectively, and we showcase the performance improvements of models trained on datasets generated by MP2D across diverse topic shift dialogue tasks.

http://arxiv.org/abs/2403.05814v1

cs.CL, cs.AI

$\textbf{S}^2$IP-LLM: Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting

Zijie Pan, Yushan Jiang, Sahil Garg, Anderson Schneider, Yuriy Nevmyvaka, Dongjin Song

Recently, there has been a growing interest in leveraging pre-trained large language models (LLMs) for various time series applications. However, the semantic space of LLMs, established through the pre-training, is still underexplored and may help yield more distinctive and informative representations to facilitate time series forecasting. To this end, we propose Semantic Space Informed Prompt learning with LLM ($S^2$IP-LLM) to align the pre-trained semantic space with time series embeddings space and perform time series forecasting based on learned prompts from the joint space. We first design a tokenization module tailored for cross-modality alignment, which explicitly concatenates patches of decomposed time series components to create embeddings that effectively encode the temporal dynamics. Next, we leverage the pre-trained word token embeddings to derive semantic anchors and align selected anchors with time series embeddings by maximizing the cosine similarity in the joint space. This way, $S^2$IP-LLM can retrieve relevant semantic anchors as prompts to provide strong indicators (context) for time series that exhibit different temporal dynamics. With thorough empirical studies on multiple benchmark datasets, we demonstrate that the proposed $S^2$IP-LLM can achieve superior forecasting performance over state-of-the-art baselines. Furthermore, our ablation studies and visualizations verify the necessity of prompt learning informed by semantic space.

http://arxiv.org/abs/2403.05798v1

cs.LG

A Novel Nuanced Conversation Evaluation Framework for Large Language Models in Mental Health

Alexander Marrapese, Basem Suleiman, Imdad Ullah, Juno Kim

Understanding the conversation abilities of Large Language Models (LLMs) can help lead to its more cautious and appropriate deployment. This is especially important for safety-critical domains like mental health, where someone's life may depend on the exact wording of a response to an urgent question. In this paper, we propose a novel framework for evaluating the nuanced conversation abilities of LLMs. Within it, we develop a series of quantitative metrics developed from literature on using psychotherapy conversation analysis literature. While we ensure that our framework and metrics are transferable by researchers to relevant adjacent domains, we apply them to the mental health field. We use our framework to evaluate several popular frontier LLMs, including some GPT and Llama models, through a verified mental health dataset. Our results show that GPT4 Turbo can perform significantly more similarly to verified therapists than other selected LLMs. We conduct additional analysis to examine how LLM conversation performance varies across specific mental health topics. Our results indicate that GPT4 Turbo performs well in achieving high correlation with verified therapists in particular topics such as Parenting and Relationships. We believe our contributions will help researchers develop better LLMs that, in turn, will more positively support people's lives.

http://arxiv.org/abs/2403.09705v1

cs.CL, cs.AI, cs.ET

A Benchmark of Domain-Adapted Large Language Models for Generating Brief Hospital Course Summaries

Asad Aali, Dave Van Veen, Yamin Ishraq Arefeen, Jason Hom, Christian Bluethgen, Eduardo Pontes Reis, Sergios Gatidis, Namuun Clifford, Joseph Daws, Arash S. Tehrani, Jangwon Kim, Akshay S. Chaudhari

Brief hospital course (BHC) summaries are common clinical documents generated by summarizing clinical notes. While large language models (LLMs) depict remarkable capabilities in automating real-world tasks, their capabilities for healthcare applications such as BHC synthesis have not been shown. To enable the adaptation of LLMs for BHC synthesis, we introduce a novel benchmark consisting of a pre-processed dataset extracted from MIMIC-IV notes, encapsulating clinical note, and brief hospital course (BHC) pairs. We assess the performance of two general-purpose LLMs and three healthcare-adapted LLMs to improve BHC synthesis from clinical notes. Using clinical notes as input for generating BHCs, we apply prompting-based (using in-context learning) and fine-tuning-based adaptation strategies to three open-source LLMs (Clinical-T5-Large, Llama2-13B, FLAN-UL2) and two proprietary LLMs (GPT-3.5, GPT-4). We quantitatively evaluate the performance of these LLMs across varying context-length inputs using conventional natural language similarity metrics. We further perform a qualitative study where five diverse clinicians blindly compare clinician-written BHCs and two LLM-generated BHCs for 30 samples across metrics of comprehensiveness, conciseness, factual correctness, and fluency. Overall, we present a new benchmark and pre-processed dataset for using LLMs in BHC synthesis from clinical notes. We observe high-quality summarization performance for both in-context proprietary and fine-tuned open-source LLMs using both quantitative metrics and a qualitative clinical reader study. We propose our work as a benchmark to motivate future works to adapt and assess the performance of LLMs in BHC synthesis.

http://arxiv.org/abs/2403.05720v1

cs.CL, cs.AI, cs.LG

SeeGULL Multilingual: a Dataset of Geo-Culturally Situated Stereotypes

Mukul Bhutani, Kevin Robinson, Vinodkumar Prabhakaran, Shachi Dave, Sunipa Dev

While generative multilingual models are rapidly being deployed, their safety and fairness evaluations are largely limited to resources collected in English. This is especially problematic for evaluations targeting inherently socio-cultural phenomena such as stereotyping, where it is important to build multi-lingual resources that reflect the stereotypes prevalent in respective language communities. However, gathering these resources, at scale, in varied languages and regions pose a significant challenge as it requires broad socio-cultural knowledge and can also be prohibitively expensive. To overcome this critical gap, we employ a recently introduced approach that couples LLM generations for scale with culturally situated validations for reliability, and build SeeGULL Multilingual, a global-scale multilingual dataset of social stereotypes, containing over 25K stereotypes, spanning 20 languages, with human annotations across 23 regions, and demonstrate its utility in identifying gaps in model evaluations. Content warning: Stereotypes shared in this paper can be offensive.

http://arxiv.org/abs/2403.05696v1

cs.CL, cs.CV

DP-TabICL: In-Context Learning with Differentially Private Tabular Data

Alycia N. Carey, Karuna Bhaila, Kennedy Edemacu, Xintao Wu

In-context learning (ICL) enables large language models (LLMs) to adapt to new tasks by conditioning on demonstrations of question-answer pairs and it has been shown to have comparable performance to costly model retraining and fine-tuning. Recently, ICL has been extended to allow tabular data to be used as demonstration examples by serializing individual records into natural language formats. However, it has been shown that LLMs can leak information contained in prompts, and since tabular data often contain sensitive information, understanding how to protect the underlying tabular data used in ICL is a critical area of research. This work serves as an initial investigation into how to use differential privacy (DP) -- the long-established gold standard for data privacy and anonymization -- to protect tabular data used in ICL. Specifically, we investigate the application of DP mechanisms for private tabular ICL via data privatization prior to serialization and prompting. We formulate two private ICL frameworks with provable privacy guarantees in both the local (LDP-TabICL) and global (GDP-TabICL) DP scenarios via injecting noise into individual records or group statistics, respectively. We evaluate our DP-based frameworks on eight real-world tabular datasets and across multiple ICL and DP settings. Our evaluations show that DP-based ICL can protect the privacy of the underlying tabular data while achieving comparable performance to non-LLM baselines, especially under high privacy regimes.

http://arxiv.org/abs/2403.05681v1

cs.CR, cs.AI, cs.LG

DeepSeek-VL: Towards Real-World Vision-Language Understanding

Haoyu Lu, Wen Liu, Bo Zhang, Bingxuan Wang, Kai Dong, Bo Liu, Jingxiang Sun, Tongzheng Ren, Zhuoshu Li, Hao Yang, Yaofeng Sun, Chengqi Deng, Hanwei Xu, Zhenda Xie, Chong Ruan

We present DeepSeek-VL, an open-source Vision-Language (VL) Model designed for real-world vision and language understanding applications. Our approach is structured around three key dimensions: We strive to ensure our data is diverse, scalable, and extensively covers real-world scenarios including web screenshots, PDFs, OCR, charts, and knowledge-based content, aiming for a comprehensive representation of practical contexts. Further, we create a use case taxonomy from real user scenarios and construct an instruction tuning dataset accordingly. The fine-tuning with this dataset substantially improves the model's user experience in practical applications. Considering efficiency and the demands of most real-world scenarios, DeepSeek-VL incorporates a hybrid vision encoder that efficiently processes high-resolution images (1024 x 1024), while maintaining a relatively low computational overhead. This design choice ensures the model's ability to capture critical semantic and detailed information across various visual tasks. We posit that a proficient Vision-Language Model should, foremost, possess strong language abilities. To ensure the preservation of LLM capabilities during pretraining, we investigate an effective VL pretraining strategy by integrating LLM training from the beginning and carefully managing the competitive dynamics observed between vision and language modalities. The DeepSeek-VL family (both 1.3B and 7B models) showcases superior user experiences as a vision-language chatbot in real-world applications, achieving state-of-the-art or competitive performance across a wide range of visual-language benchmarks at the same model size while maintaining robust performance on language-centric benchmarks. We have made both 1.3B and 7B models publicly accessible to foster innovations based on this foundation model.

http://arxiv.org/abs/2403.05525v2

cs.AI

Beyond Finite Data: Towards Data-free Out-of-distribution Generalization via Extrapolation

Yijiang Li, Sucheng Ren, Weipeng Deng, Yuzhi Xu, Ying Gao, Edith Ngai, Haohan Wang

Out-of-distribution (OOD) generalization is a favorable yet challenging property for deep neural networks. The core challenges lie in the limited availability of source domains that help models learn an invariant representation from the spurious features. Various domain augmentation have been proposed but largely rely on interpolating existing domains and frequently face difficulties in creating truly "novel" domains. Humans, on the other hand, can easily extrapolate novel domains, thus, an intriguing question arises: How can neural networks extrapolate like humans and achieve OOD generalization? We introduce a novel approach to domain extrapolation that leverages reasoning ability and the extensive knowledge encapsulated within large language models (LLMs) to synthesize entirely new domains. Starting with the class of interest, we query the LLMs to extract relevant knowledge for these novel domains. We then bridge the gap between the text-centric knowledge derived from LLMs and the pixel input space of the model using text-to-image generation techniques. By augmenting the training set of domain generalization datasets with high-fidelity, photo-realistic images of these new domains, we achieve significant improvements over all existing methods, as demonstrated in both single and multi-domain generalization across various benchmarks. With the ability to extrapolate any domains for any class, our method has the potential to learn a generalized model for any task without any data. To illustrate, we put forth a much more difficult setting termed, data-free domain generalization, that aims to learn a generalized model in the absence of any collected data. Our empirical findings support the above argument and our methods exhibit commendable performance in this setting, even surpassing the supervised setting by approximately 1-2\% on datasets such as VLCS.

http://arxiv.org/abs/2403.05523v2

cs.CV

Cost-Performance Optimization for Processing Low-Resource Language Tasks Using Commercial LLMs

Arijit Nag, Animesh Mukherjee, Niloy Ganguly, Soumen Chakrabarti

Large Language Models (LLMs) exhibit impressive zero/few-shot inference and generation quality for high-resource languages (HRLs). A few of them have been trained on low-resource languages (LRLs) and give decent performance. Owing to the prohibitive costs of training LLMs, they are usually used as a network service, with the client charged by the count of input and output tokens. The number of tokens strongly depends on the script and language, as well as the LLM's subword vocabulary. We show that LRLs are at a pricing disadvantage, because the well-known LLMs produce more tokens for LRLs than HRLs. This is because most currently popular LLMs are optimized for HRL vocabularies. Our objective is to level the playing field: reduce the cost of processing LRLs in contemporary LLMs while ensuring that predictive and generative qualities are not compromised. As means to reduce the number of tokens processed by the LLM, we consider code-mixing, translation, and transliteration of LRLs to HRLs. We perform an extensive study using the IndicXTREME classification and six generative tasks dataset, covering 15 Indic and 3 other languages, while using GPT-4 (one of the costliest LLM services released so far) as a commercial LLM. We observe and analyze interesting patterns involving token count, cost, and quality across a multitude of languages and tasks. We show that choosing the best policy to interact with the LLM can reduce cost by 90% while giving better or comparable performance compared to communicating with the LLM in the original LRL.

http://arxiv.org/abs/2403.05434v2

cs.CL

ChatASU: Evoking LLM's Reflexion to Truly Understand Aspect Sentiment in Dialogues

Yiding Liu, Jingjing Wang, Jiamin Luo, Tao Zeng, Guodong Zhou

Aspect Sentiment Understanding (ASU) in interactive scenarios (e.g., Question-Answering and Dialogue) has attracted ever-more interest in recent years and achieved important progresses. However, existing studies on interactive ASU largely ignore the coreference issue for opinion targets (i.e., aspects), while this phenomenon is ubiquitous in interactive scenarios especially dialogues, limiting the ASU performance. Recently, large language models (LLMs) shows the powerful ability to integrate various NLP tasks with the chat paradigm. In this way, this paper proposes a new Chat-based Aspect Sentiment Understanding (ChatASU) task, aiming to explore LLMs' ability in understanding aspect sentiments in dialogue scenarios. Particularly, this ChatASU task introduces a sub-task, i.e., Aspect Chain Reasoning (ACR) task, to address the aspect coreference issue. On this basis, we propose a Trusted Self-reflexion Approach (TSA) with ChatGLM as backbone to ChatASU. Specifically, this TSA treats the ACR task as an auxiliary task to boost the performance of the primary ASU task, and further integrates trusted learning into reflexion mechanisms to alleviate the LLMs-intrinsic factual hallucination problem in TSA. Furthermore, a high-quality ChatASU dataset is annotated to evaluate TSA, and extensive experiments show that our proposed TSA can significantly outperform several state-of-the-art baselines, justifying the effectiveness of TSA to ChatASU and the importance of considering the coreference and hallucination issues in ChatASU.

http://arxiv.org/abs/2403.05326v4

cs.CL, cs.AI

ACLSum: A New Dataset for Aspect-based Summarization of Scientific Publications

Sotaro Takeshita, Tommaso Green, Ines Reinig, Kai Eckert, Simone Paolo Ponzetto

Extensive efforts in the past have been directed toward the development of summarization datasets. However, a predominant number of these resources have been (semi)-automatically generated, typically through web data crawling, resulting in subpar resources for training and evaluating summarization systems, a quality compromise that is arguably due to the substantial costs associated with generating ground-truth summaries, particularly for diverse languages and specialized domains. To address this issue, we present ACLSum, a novel summarization dataset carefully crafted and evaluated by domain experts. In contrast to previous datasets, ACLSum facilitates multi-aspect summarization of scientific papers, covering challenges, approaches, and outcomes in depth. Through extensive experiments, we evaluate the quality of our resource and the performance of models based on pretrained language models and state-of-the-art large language models (LLMs). Additionally, we explore the effectiveness of extractive versus abstractive summarization within the scholarly domain on the basis of automatically discovered aspects. Our results corroborate previous findings in the general domain and indicate the general superiority of end-to-end aspect-based summarization. Our data is released at https://github.com/sobamchan/aclsum.

http://arxiv.org/abs/2403.05303v1

cs.CL

LLM4Decompile: Decompiling Binary Code with Large Language Models

Hanzhuo Tan, Qi Luo, Jing Li, Yuqun Zhang

Decompilation aims to restore compiled code to human-readable source code, but struggles with details like names and structure. Large language models (LLMs) show promise for programming tasks, motivating their application to decompilation. However, there does not exist any open-source LLM for decompilation. Moreover, existing decompilation evaluation systems mainly consider token-level accuracy and largely ignore code executability, which is the most important feature of any program. Therefore, we release the first open-access decompilation LLMs ranging from 1B to 33B pre-trained on 4 billion tokens of C source code and the corresponding assembly code. The open-source LLMs can serve as baselines for further development in the field. To ensure practical program evaluation, we introduce Decompile-Eval, the first dataset that considers re-compilability and re-executability for decompilation. The benchmark emphasizes the importance of evaluating the decompilation model from the perspective of program semantics. Experiments indicate that our LLM4Decompile has demonstrated the capability to accurately decompile 21% of the assembly code, which achieves a 50% improvement over GPT-4. Our code, dataset, and models are released at https://github.com/albertan017/LLM4Decompile

http://arxiv.org/abs/2403.05286v1

cs.PL, cs.CL

Generator-Guided Crowd Reaction Assessment

Sohom Ghosh, Chung-Chi Chen, Sudip Kumar Naskar

In the realm of social media, understanding and predicting post reach is a significant challenge. This paper presents a Crowd Reaction AssessMent (CReAM) task designed to estimate if a given social media post will receive more reaction than another, a particularly essential task for digital marketers and content writers. We introduce the Crowd Reaction Estimation Dataset (CRED), consisting of pairs of tweets from The White House with comparative measures of retweet count. The proposed Generator-Guided Estimation Approach (GGEA) leverages generative Large Language Models (LLMs), such as ChatGPT, FLAN-UL2, and Claude, to guide classification models for making better predictions. Our results reveal that a fine-tuned FLANG-RoBERTa model, utilizing a cross-encoder architecture with tweet content and responses generated by Claude, performs optimally. We further use a T5-based paraphraser to generate paraphrases of a given post and demonstrate GGEA's ability to predict which post will elicit the most reactions. We believe this novel application of LLMs provides a significant advancement in predicting social media post reach.

http://arxiv.org/abs/2403.09702v1

cs.CL, I.2.7

Overcoming Reward Overoptimization via Adversarial Policy Optimization with Lightweight Uncertainty Estimation

Xiaoying Zhang, Jean-Francois Ton, Wei Shen, Hongning Wang, Yang Liu

We introduce Adversarial Policy Optimization (AdvPO), a novel solution to the pervasive issue of reward over-optimization in Reinforcement Learning from Human Feedback (RLHF) for Large Language Models (LLMs). Over-optimization occurs when a reward model serves as an imperfect proxy for human preference, and RL-driven policy optimization erroneously exploits reward inaccuracies. In this paper, we begin by introducing a lightweight way to quantify uncertainties in rewards, relying solely on the last layer embeddings of the reward model, without the need for computationally expensive reward ensembles. AdvPO then addresses a distributionally robust optimization problem centred around the confidence interval of the reward model's predictions for policy improvement. Through comprehensive experiments on the Anthropic HH and TL;DR summarization datasets, we illustrate the efficacy of AdvPO in mitigating the overoptimization issue, consequently resulting in enhanced performance as evaluated through human-assisted evaluation.

http://arxiv.org/abs/2403.05171v1

cs.LG, cs.AI

Benchmarking Large Language Models for Molecule Prediction Tasks

Zhiqiang Zhong, Kuangyu Zhou, Davide Mottin

Large Language Models (LLMs) stand at the forefront of a number of Natural Language Processing (NLP) tasks. Despite the widespread adoption of LLMs in NLP, much of their potential in broader fields remains largely unexplored, and significant limitations persist in their design and implementation. Notably, LLMs struggle with structured data, such as graphs, and often falter when tasked with answering domain-specific questions requiring deep expertise, such as those in biology and chemistry. In this paper, we explore a fundamental question: Can LLMs effectively handle molecule prediction tasks? Rather than pursuing top-tier performance, our goal is to assess how LLMs can contribute to diverse molecule tasks. We identify several classification and regression prediction tasks across six standard molecule datasets. Subsequently, we carefully design a set of prompts to query LLMs on these tasks and compare their performance with existing Machine Learning (ML) models, which include text-based models and those specifically designed for analysing the geometric structure of molecules. Our investigation reveals several key insights: Firstly, LLMs generally lag behind ML models in achieving competitive performance on molecule tasks, particularly when compared to models adept at capturing the geometric structure of molecules, highlighting the constrained ability of LLMs to comprehend graph data. Secondly, LLMs show promise in enhancing the performance of ML models when used collaboratively. Lastly, we engage in a discourse regarding the challenges and promising avenues to harness LLMs for molecule prediction tasks. The code and models are available at https://github.com/zhiqiangzhongddu/LLMaMol.

http://arxiv.org/abs/2403.05075v1

cs.LG, q-bio.BM

Can we obtain significant success in RST discourse parsing by using Large Language Models?

Aru Maekawa, Tsutomu Hirao, Hidetaka Kamigaito, Manabu Okumura

Recently, decoder-only pre-trained large language models (LLMs), with several tens of billion parameters, have significantly impacted a wide range of natural language processing (NLP) tasks. While encoder-only or encoder-decoder pre-trained language models have already proved to be effective in discourse parsing, the extent to which LLMs can perform this task remains an open research question. Therefore, this paper explores how beneficial such LLMs are for Rhetorical Structure Theory (RST) discourse parsing. Here, the parsing process for both fundamental top-down and bottom-up strategies is converted into prompts, which LLMs can work with. We employ Llama 2 and fine-tune it with QLoRA, which has fewer parameters that can be tuned. Experimental results on three benchmark datasets, RST-DT, Instr-DT, and the GUM corpus, demonstrate that Llama 2 with 70 billion parameters in the bottom-up strategy obtained state-of-the-art (SOTA) results with significant differences. Furthermore, our parsers demonstrated generalizability when evaluated on RST-DT, showing that, in spite of being trained with the GUM corpus, it obtained similar performances to those of existing parsers trained with RST-DT.

http://arxiv.org/abs/2403.05065v1

cs.CL

Aligning Large Language Models for Controllable Recommendations

Wensheng Lu, Jianxun Lian, Wei Zhang, Guanghua Li, Mingyang Zhou, Hao Liao, Xing Xie

Inspired by the exceptional general intelligence of Large Language Models (LLMs), researchers have begun to explore their application in pioneering the next generation of recommender systems - systems that are conversational, explainable, and controllable. However, existing literature primarily concentrates on integrating domain-specific knowledge into LLMs to enhance accuracy, often neglecting the ability to follow instructions. To address this gap, we initially introduce a collection of supervised learning tasks, augmented with labels derived from a conventional recommender model, aimed at explicitly improving LLMs' proficiency in adhering to recommendation-specific instructions. Subsequently, we develop a reinforcement learning-based alignment procedure to further strengthen LLMs' aptitude in responding to users' intentions and mitigating formatting errors. Through extensive experiments on two real-world datasets, our method markedly advances the capability of LLMs to comply with instructions within recommender systems, while sustaining a high level of accuracy performance.

http://arxiv.org/abs/2403.05063v1

cs.IR, cs.AI, 68T50

DiffChat: Learning to Chat with Text-to-Image Synthesis Models for Interactive Image Creation

Jiapeng Wang, Chengyu Wang, Tingfeng Cao, Jun Huang, Lianwen Jin

We present DiffChat, a novel method to align Large Language Models (LLMs) to "chat" with prompt-as-input Text-to-Image Synthesis (TIS) models (e.g., Stable Diffusion) for interactive image creation. Given a raw prompt/image and a user-specified instruction, DiffChat can effectively make appropriate modifications and generate the target prompt, which can be leveraged to create the target image of high quality. To achieve this, we first collect an instruction-following prompt engineering dataset named InstructPE for the supervised training of DiffChat. Next, we propose a reinforcement learning framework with the feedback of three core criteria for image creation, i.e., aesthetics, user preference, and content integrity. It involves an action-space dynamic modification technique to obtain more relevant positive samples and harder negative samples during the off-policy sampling. Content integrity is also introduced into the value estimation function for further improvement of produced images. Our method can exhibit superior performance than baseline models and strong competitors based on both automatic and human evaluations, which fully demonstrates its effectiveness.

http://arxiv.org/abs/2403.04997v1

cs.CL, cs.CV

Electrocardiogram Instruction Tuning for Report Generation

Zhongwei Wan, Che Liu, Xin Wang, Chaofan Tao, Hui Shen, Zhenwu Peng, Jie Fu, Rossella Arcucci, Huaxiu Yao, Mi Zhang

Electrocardiogram (ECG) serves as the primary non-invasive diagnostic tool for cardiac conditions monitoring, are crucial in assisting clinicians. Recent studies have concentrated on classifying cardiac conditions using ECG data but have overlooked ECG report generation, which is not only time-consuming but also requires clinical expertise. To automate ECG report generation and ensure its versatility, we propose the Multimodal ECG Instruction Tuning (MEIT) framework, the \textit{first} attempt to tackle ECG report generation with LLMs and multimodal instructions. To facilitate future research, we establish a benchmark to evaluate MEIT with various LLMs backbones across two large-scale ECG datasets. Our approach uniquely aligns the representations of the ECG signal and the report, and we conduct extensive experiments to benchmark MEIT with nine open source LLMs, using more than 800,000 ECG reports. MEIT's results underscore the superior performance of instruction-tuned LLMs, showcasing their proficiency in quality report generation, zero-shot capabilities, and resilience to signal perturbation. These findings emphasize the efficacy of our MEIT framework and its potential for real-world clinical application.

http://arxiv.org/abs/2403.04945v2

cs.CL, cs.LG, eess.SP

$\text{R}^2$-Bench: Benchmarking the Robustness of Referring Perception Models under Perturbations

Xiang Li, Kai Qiu, Jinglu Wang, Xiaohao Xu, Rita Singh, Kashu Yamazak, Hao Chen, Xiaonan Huang, Bhiksha Raj

Referring perception, which aims at grounding visual objects with multimodal referring guidance, is essential for bridging the gap between humans, who provide instructions, and the environment where intelligent systems perceive. Despite progress in this field, the robustness of referring perception models (RPMs) against disruptive perturbations is not well explored. This work thoroughly assesses the resilience of RPMs against various perturbations in both general and specific contexts. Recognizing the complex nature of referring perception tasks, we present a comprehensive taxonomy of perturbations, and then develop a versatile toolbox for synthesizing and evaluating the effects of composite disturbances. Employing this toolbox, we construct $\text{R}^2$-Bench, a benchmark for assessing the Robustness of Referring perception models under noisy conditions across five key tasks. Moreover, we propose the $\text{R}^2$-Agent, an LLM-based agent that simplifies and automates model evaluation via natural language instructions. Our investigation uncovers the vulnerabilities of current RPMs to various perturbations and provides tools for assessing model robustness, potentially promoting the safe and resilient integration of intelligent systems into complex real-world scenarios.

http://arxiv.org/abs/2403.04924v1

cs.CV

ConstitutionalExperts: Training a Mixture of Principle-based Prompts

Savvas Petridis, Ben Wedin, Ann Yuan, James Wexler, Nithum Thain

Large language models (LLMs) are highly capable at a variety of tasks given the right prompt, but writing one is still a difficult and tedious process. In this work, we introduce ConstitutionalExperts, a method for learning a prompt consisting of constitutional principles (i.e. rules), given a training dataset. Unlike prior methods that optimize the prompt as a single entity, our method incrementally improves the prompt by surgically editing individual principles. We also show that we can improve overall performance by learning unique prompts for different semantic regions of the training data and using a mixture-of-experts (MoE) architecture to route inputs at inference time. We compare our method to other state of the art prompt-optimization techniques across six benchmark datasets. We also investigate whether MoE improves these other techniques. Our results suggest that ConstitutionalExperts outperforms other prompt optimization techniques by 10.9% (F1) and that mixture-of-experts improves all techniques, suggesting its broad applicability.

http://arxiv.org/abs/2403.04894v1

cs.CL, cs.AI

Few shot chain-of-thought driven reasoning to prompt LLMs for open ended medical question answering

Ojas Gramopadhye, Saeel Sandeep Nachane, Prateek Chanda, Ganesh Ramakrishnan, Kshitij Sharad Jadhav, Yatin Nandwani, Dinesh Raghu, Sachindra Joshi

Large Language models (LLMs) have demonstrated significant potential in transforming healthcare by automating tasks such as clinical documentation, information retrieval, and decision support. In this aspect, carefully engineered prompts have emerged as a powerful tool for using LLMs for medical scenarios, e.g., patient clinical scenarios. In this paper, we propose a modified version of the MedQA-USMLE dataset, which is subjective, to mimic real-life clinical scenarios. We explore the Chain of Thought (CoT) reasoning based on subjective response generation for the modified MedQA-USMLE dataset with appropriate LM-driven forward reasoning for correct responses to the medical questions. Keeping in mind the importance of response verification in the medical setting, we utilize a reward training mechanism whereby the language model also provides an appropriate verified response for a particular response to a clinical question. In this regard, we also include human-in-the-loop for different evaluation aspects. We develop better in-contrast learning strategies by modifying the 5-shot-codex-CoT-prompt from arXiv:2207.08143 for the subjective MedQA dataset and developing our incremental-reasoning prompt. Our evaluations show that the incremental reasoning prompt performs better than the modified codex prompt in certain scenarios. We also show that greedy decoding with the incremental reasoning method performs better than other strategies, such as prompt chaining and eliminative reasoning.

http://arxiv.org/abs/2403.04890v1

cs.CL

SnapNTell: Enhancing Entity-Centric Visual Question Answering with Retrieval Augmented Multimodal LLM

Jielin Qiu, Andrea Madotto, Zhaojiang Lin, Paul A. Crook, Yifan Ethan Xu, Xin Luna Dong, Christos Faloutsos, Lei Li, Babak Damavandi, Seungwhan Moon

Vision-extended LLMs have made significant strides in Visual Question Answering (VQA). Despite these advancements, VLLMs still encounter substantial difficulties in handling queries involving long-tail entities, with a tendency to produce erroneous or hallucinated responses. In this work, we introduce a novel evaluative benchmark named \textbf{SnapNTell}, specifically tailored for entity-centric VQA. This task aims to test the models' capabilities in identifying entities and providing detailed, entity-specific knowledge. We have developed the \textbf{SnapNTell Dataset}, distinct from traditional VQA datasets: (1) It encompasses a wide range of categorized entities, each represented by images and explicitly named in the answers; (2) It features QA pairs that require extensive knowledge for accurate responses. The dataset is organized into 22 major categories, containing 7,568 unique entities in total. For each entity, we curated 10 illustrative images and crafted 10 knowledge-intensive QA pairs. To address this novel task, we devised a scalable, efficient, and transparent retrieval-augmented multimodal LLM. Our approach markedly outperforms existing methods on the SnapNTell dataset, achieving a 66.5\% improvement in the BELURT score. We will soon make the dataset and the source code publicly accessible.

http://arxiv.org/abs/2403.04735v1

cs.CV

How Far Are We from Intelligent Visual Deductive Reasoning?

Yizhe Zhang, He Bai, Ruixiang Zhang, Jiatao Gu, Shuangfei Zhai, Josh Susskind, Navdeep Jaitly

Vision-Language Models (VLMs) such as GPT-4V have recently demonstrated incredible strides on diverse vision language tasks. We dig into vision-based deductive reasoning, a more sophisticated but less explored realm, and find previously unexposed blindspots in the current SOTA VLMs. Specifically, we leverage Raven's Progressive Matrices (RPMs), to assess VLMs' abilities to perform multi-hop relational and deductive reasoning relying solely on visual clues. We perform comprehensive evaluations of several popular VLMs employing standard strategies such as in-context learning, self-consistency, and Chain-of-thoughts (CoT) on three diverse datasets, including the Mensa IQ test, IntelligenceTest, and RAVEN. The results reveal that despite the impressive capabilities of LLMs in text-based reasoning, we are still far from achieving comparable proficiency in visual deductive reasoning. We found that certain standard strategies that are effective when applied to LLMs do not seamlessly translate to the challenges presented by visual reasoning tasks. Moreover, a detailed analysis reveals that VLMs struggle to solve these tasks mainly because they are unable to perceive and comprehend multiple, confounding abstract patterns in RPM examples.

http://arxiv.org/abs/2403.04732v2

cs.AI, cs.CL, cs.CV

Wiki-TabNER:Advancing Table Interpretation Through Named Entity Recognition

Aneta Koleva, Martin Ringsquandl, Ahmed Hatem, Thomas Runkler, Volker Tresp

Web tables contain a large amount of valuable knowledge and have inspired tabular language models aimed at tackling table interpretation (TI) tasks. In this paper, we analyse a widely used benchmark dataset for evaluation of TI tasks, particularly focusing on the entity linking task. Our analysis reveals that this dataset is overly simplified, potentially reducing its effectiveness for thorough evaluation and failing to accurately represent tables as they appear in the real-world. To overcome this drawback, we construct and annotate a new more challenging dataset. In addition to introducing the new dataset, we also introduce a novel problem aimed at addressing the entity linking task: named entity recognition within cells. Finally, we propose a prompting framework for evaluating the newly developed large language models (LLMs) on this novel TI task. We conduct experiments on prompting LLMs under various settings, where we use both random and similarity-based selection to choose the examples presented to the models. Our ablation study helps us gain insights into the impact of the few-shot examples. Additionally, we perform qualitative analysis to gain insights into the challenges encountered by the models and to understand the limitations of the proposed dataset.

http://arxiv.org/abs/2403.04577v1

cs.AI, cs.CL

Do Large Language Model Understand Multi-Intent Spoken Language ?

Shangjian Yin, Peijie Huang, Yuhong Xu, Haojing Huang, Jiatian Chen

This research signifies a considerable breakthrough in leveraging Large Language Models (LLMs) for multi-intent spoken language understanding (SLU). Our approach re-imagines the use of entity slots in multi-intent SLU applications, making the most of the generative potential of LLMs within the SLU landscape, leading to the development of the EN-LLM series. Furthermore, we introduce the concept of Sub-Intent Instruction (SII) to amplify the analysis and interpretation of complex, multi-intent communications, which further supports the creation of the ENSI-LLM models series. Our novel datasets, identified as LM-MixATIS and LM-MixSNIPS, are synthesized from existing benchmarks. The study evidences that LLMs may match or even surpass the performance of the current best multi-intent SLU models. We also scrutinize the performance of LLMs across a spectrum of intent configurations and dataset distributions. On top of this, we present two revolutionary metrics - Entity Slot Accuracy (ESA) and Combined Semantic Accuracy (CSA) - to facilitate a detailed assessment of LLM competence in this multifaceted field." Our code and datasets are available at \url{https://github.com/SJY8460/SLM}.

http://arxiv.org/abs/2403.04481v3

cs.CL, cs.AI

Pearl: A Review-driven Persona-Knowledge Grounded Conversational Recommendation Dataset

Minjin Kim, Minju Kim, Hana Kim, Beong-woo Kwak, Soyeon Chun, Hyunseo Kim, SeongKu Kang, Youngjae Yu, Jinyoung Yeo, Dongha Lee

Conversational recommender system is an emerging area that has garnered an increasing interest in the community, especially with the advancements in large language models (LLMs) that enable diverse reasoning over conversational input. Despite the progress, the field has many aspects left to explore. The currently available public datasets for conversational recommendation lack specific user preferences and explanations for recommendations, hindering high-quality recommendations. To address such challenges, we present a novel conversational recommendation dataset named PEARL, synthesized with persona- and knowledge-augmented LLM simulators. We obtain detailed persona and knowledge from real-world reviews and construct a large-scale dataset with over 57k dialogues. Our experimental results demonstrate that utterances in PEARL include more specific user preferences, show expertise in the target domain, and provide recommendations more relevant to the dialogue context than those in prior datasets.

http://arxiv.org/abs/2403.04460v3

cs.CL

Low-Resource Court Judgment Summarization for Common Law Systems

Shuaiqi Liu, Jiannong Cao, Yicong Li, Ruosong Yang, Zhiyuan Wen

Common law courts need to refer to similar precedents' judgments to inform their current decisions. Generating high-quality summaries of court judgment documents can facilitate legal practitioners to efficiently review previous cases and assist the general public in accessing how the courts operate and how the law is applied. Previous court judgment summarization research focuses on civil law or a particular jurisdiction's judgments. However, judges can refer to the judgments from all common law jurisdictions. Current summarization datasets are insufficient to satisfy the demands of summarizing precedents across multiple jurisdictions, especially when labeled data are scarce for many jurisdictions. To address the lack of datasets, we present CLSum, the first dataset for summarizing multi-jurisdictional common law court judgment documents. Besides, this is the first court judgment summarization work adopting large language models (LLMs) in data augmentation, summary generation, and evaluation. Specifically, we design an LLM-based data augmentation method incorporating legal knowledge. We also propose a legal knowledge enhanced evaluation metric based on LLM to assess the quality of generated judgment summaries. Our experimental results verify that the LLM-based summarization methods can perform well in the few-shot and zero-shot settings. Our LLM-based data augmentation method can mitigate the impact of low data resources. Furthermore, we carry out comprehensive comparative experiments to find essential model components and settings that are capable of enhancing summarization performance.

http://arxiv.org/abs/2403.04454v1

cs.CL, cs.AI, I.2.7; I.7

HaluEval-Wild: Evaluating Hallucinations of Language Models in the Wild

Zhiying Zhu, Yiming Yang, Zhiqing Sun

Hallucinations pose a significant challenge to the reliability of large language models (LLMs) in critical domains. Recent benchmarks designed to assess LLM hallucinations within conventional NLP tasks, such as knowledge-intensive question answering (QA) and summarization, are insufficient for capturing the complexities of user-LLM interactions in dynamic, real-world settings. To address this gap, we introduce HaluEval-Wild, the first benchmark specifically designed to evaluate LLM hallucinations in the wild. We meticulously collect challenging (adversarially filtered by Alpaca) user queries from existing real-world user-LLM interaction datasets, including ShareGPT, to evaluate the hallucination rates of various LLMs. Upon analyzing the collected queries, we categorize them into five distinct types, which enables a fine-grained analysis of the types of hallucinations LLMs exhibit, and synthesize the reference answers with the powerful GPT-4 model and retrieval-augmented generation (RAG). Our benchmark offers a novel approach towards enhancing our comprehension and improvement of LLM reliability in scenarios reflective of real-world interactions. Our benchmark is available at https://github.com/Dianezzy/HaluEval-Wild.

http://arxiv.org/abs/2403.04307v2

cs.CL