Authors: Hyunkyung Han, Jaesik Choi
Abstract: Biomedical text summarization is a critical tool that enables clinicians to effectively ascertain patient status. Traditionally, text summarization has been accomplished with transformer models, which are capable of compressing long documents into brief summaries. However, transformer models are known to be among the most challenging natural language processing (NLP) tasks. Specifically, GPT models have a tendency to generate factual errors, lack context, and oversimplify words. To address these limitations, we replaced the attention mechanism in the GPT model with a pointer network. This modification was designed to preserve the core values of the original text during the summarization process. The effectiveness of the Pointer-GPT model was evaluated using the ROUGE score. The results demonstrated that Pointer-GPT outperformed the original GPT model. These findings suggest that pointer networks can be a valuable addition to EMR systems and can provide clinicians with more accurate and informative summaries of patient medical records. This research has the potential to usher in a new paradigm in EMR systems and to revolutionize the way that clinicians interact with patient medical records.
Authors: Sourya Dipta Das, Yash Vadi, Kuldeep Yadav
Abstract: Automated Essay Scoring (AES) systems are widely popular in the market as they constitute a cost-effective and time-effective option for grading systems. Nevertheless, many studies have demonstrated that the AES system fails to assign lower grades to irrelevant responses. Thus, detecting the off-topic response in automated essay scoring is crucial in practical tasks where candidates write unrelated text responses to the given task in the question. In this paper, we are proposing an unsupervised technique that jointly scores essays and detects off-topic essays. The proposed Automated Open Essay Scoring (AOES) model uses a novel topic regularization module (TRM), which can be attached on top of a transformer model, and is trained using a proposed hybrid loss function. After training, the AOES model is further used to calculate the Mahalanobis distance score for off-topic essay detection. Our proposed method outperforms the baseline we created and earlier conventional methods on two essay-scoring datasets in off-topic detection as well as on-topic scoring. Experimental evaluation results on different adversarial strategies also show how the suggested method is robust for detecting possible human-level perturbations.
Authors: Shafiuddin Rehan Ahmed, George Arthur Baker, Evi Judge, Michael Regan, Kristin Wright-Bettner, Martha Palmer, James H. Martin
Abstract: Event Coreference Resolution (ECR) as a pairwise mention classification task is expensive both for automated systems and manual annotations. The task's quadratic difficulty is exacerbated when using Large Language Models (LLMs), making prompt engineering for ECR prohibitively costly. In this work, we propose a graphical representation of events, X-AMR, anchored around individual mentions using a \textbf{cross}-document version of \textbf{A}bstract \textbf{M}eaning \textbf{R}epresentation. We then linearize the ECR with a novel multi-hop coreference algorithm over the event graphs. The event graphs simplify ECR, making it a) LLM cost-effective, b) compositional and interpretable, and c) easily annotated. For a fair assessment, we first enrich an existing ECR benchmark dataset with these event graphs using an annotator-friendly tool we introduce. Then, we employ GPT-4, the newest LLM by OpenAI, for these annotations. Finally, using the ECR algorithm, we assess GPT-4 against humans and analyze its limitations. Through this research, we aim to advance the state-of-the-art for efficient ECR and shed light on the potential shortcomings of current LLMs at this task. Code and annotations: \url{https://github.com/ahmeshaf/gpt_coref}
Authors: Fan Zhou
Abstract: This study explores the distinctions between neural machine translation (NMT) and human translation (HT) through the lens of translation relations. It benchmarks HT to assess the translation techniques produced by an NMT system and aims to address three key research questions: the differences in overall translation relations between NMT and HT, how each utilizes non-literal translation techniques, and the variations in factors influencing their use of specific non-literal techniques. The research employs two parallel corpora, each spanning nine genres with the same source texts with one translated by NMT and the other by humans. Translation relations in these corpora are manually annotated on aligned pairs, enabling a comparative analysis that draws on linguistic insights, including semantic and syntactic nuances such as hypernyms and alterations in part-of-speech tagging. The results indicate that NMT relies on literal translation significantly more than HT across genres. While NMT performs comparably to HT in employing syntactic non-literal translation techniques, it falls behind in semantic-level performance.
Authors: Jesse Atuhurra, Hidetaka Kamigaito
Abstract: Natural language processing (NLP) has grown significantly since the advent of the Transformer architecture. Transformers have given birth to pre-trained large language models (PLMs). There has been tremendous improvement in the performance of NLP systems across several tasks. NLP systems are on par or, in some cases, better than humans at accomplishing specific tasks. However, it remains the norm that \emph{better quality datasets at the time of pretraining enable PLMs to achieve better performance, regardless of the task.} The need to have quality datasets has prompted NLP researchers to continue creating new datasets to satisfy particular needs. For example, the two top NLP conferences, ACL and EMNLP, accepted ninety-two papers in 2022, introducing new datasets. This work aims to uncover the trends and insights mined within these datasets. Moreover, we provide valuable suggestions to researchers interested in curating datasets in the future.
Authors: Javier J. Sanchez-Medina
Abstract: After the launch of ChatGPT v.4 there has been a global vivid discussion on the ability of this artificial intelligence powered platform and some other similar ones for the automatic production of all kinds of texts, including scientific and technical texts. This has triggered a reflection in many institutions on whether education and academic procedures should be adapted to the fact that in future many texts we read will not be written by humans (students, scholars, etc.), at least, not entirely. In this work it is proposed a new methodology to classify texts coming from an automatic text production engine or a human, based on Sentiment Analysis as a source for feature engineering independent variables and then train with them a Random Forest classification algorithm. Using four different sentiment lexicons, a number of new features where produced, and then fed to a machine learning random forest methodology, to train such a model. Results seem very convincing that this may be a promising research line to detect fraud, in such environments where human are supposed to be the source of texts.
Authors: Zhuochun Li, Bo Xie, Robin Hilsabeck, Alyssa Aguirre, Ning Zou, Zhimeng Luo, Daqing He
Abstract: Evidence suggests that different prompts lead large language models (LLMs) to generate responses with varying quality. Yet, little is known about prompts' effects on response quality in healthcare domains. In this exploratory study, we address this gap, focusing on a specific healthcare domain: dementia caregiving. We first developed an innovative prompt template with three components: (1) system prompts (SPs) featuring 4 different roles; (2) an initialization prompt; and (3) task prompts (TPs) specifying different levels of details, totaling 12 prompt combinations. Next, we selected 3 social media posts containing complicated, real-world questions about dementia caregivers' challenges in 3 areas: memory loss and confusion, aggression, and driving. We then entered these posts into GPT-4, with our 12 prompts, to generate 12 responses per post, totaling 36 responses. We compared the word count of the 36 responses to explore potential differences in response length. Two experienced dementia care clinicians on our team assessed the response quality using a rating scale with 5 quality indicators: factual, interpretation, application, synthesis, and comprehensiveness (scoring range: 0-5; higher scores indicate higher quality).
Authors: Simone Tedeschi, Felix Friedrich, Patrick Schramowski, Kristian Kersting, Roberto Navigli, Huu Nguyen, Bo Li
Abstract: When building Large Language Models (LLMs), it is paramount to bear safety in mind and protect them with guardrails. Indeed, LLMs should never generate content promoting or normalizing harmful, illegal, or unethical behavior that may contribute to harm to individuals or society. This principle applies to both normal and adversarial use. In response, we introduce ALERT, a large-scale benchmark to assess safety based on a novel fine-grained risk taxonomy. It is designed to evaluate the safety of LLMs through red teaming methodologies and consists of more than 45k instructions categorized using our novel taxonomy. By subjecting LLMs to adversarial testing scenarios, ALERT aims to identify vulnerabilities, inform improvements, and enhance the overall safety of the language models. Furthermore, the fine-grained taxonomy enables researchers to perform an in-depth evaluation that also helps one to assess the alignment with various policies. In our experiments, we extensively evaluate 10 popular open- and closed-source LLMs and demonstrate that many of them still struggle to attain reasonable levels of safety.
Authors: Andi Zhang, Tim Z. Xiao, Weiyang Liu, Robert Bamler, Damon Wischik
Abstract: We revisit the likelihood ratio between a pretrained large language model (LLM) and its finetuned variant as a criterion for out-of-distribution (OOD) detection. The intuition behind such a criterion is that, the pretrained LLM has the prior knowledge about OOD data due to its large amount of training data, and once finetuned with the in-distribution data, the LLM has sufficient knowledge to distinguish their difference. Leveraging the power of LLMs, we show that, for the first time, the likelihood ratio can serve as an effective OOD detector. Moreover, we apply the proposed LLM-based likelihood ratio to detect OOD questions in question-answering (QA) systems, which can be used to improve the performance of specialized LLMs for general questions. Given that likelihood can be easily obtained by the loss functions within contemporary neural network frameworks, it is straightforward to implement this approach in practice. Since both the pretrained LLMs and its various finetuned models are available, our proposed criterion can be effortlessly incorporated for OOD detection without the need for further training. We conduct comprehensive evaluation across on multiple settings, including far OOD, near OOD, spam detection, and QA scenarios, to demonstrate the effectiveness of the method.
Authors: Teo Susnjak, Peter Hwang, Napoleon H. Reyes, Andre L. C. Barczak, Timothy R. McIntosh, Surangika Ranathunga
Abstract: This research pioneers the use of fine-tuned Large Language Models (LLMs) to automate Systematic Literature Reviews (SLRs), presenting a significant and novel contribution in integrating AI to enhance academic research methodologies. Our study employed the latest fine-tuning methodologies together with open-sourced LLMs, and demonstrated a practical and efficient approach to automating the final execution stages of an SLR process that involves knowledge synthesis. The results maintained high fidelity in factual accuracy in LLM responses, and were validated through the replication of an existing PRISMA-conforming SLR. Our research proposed solutions for mitigating LLM hallucination and proposed mechanisms for tracking LLM responses to their sources of information, thus demonstrating how this approach can meet the rigorous demands of scholarly research. The findings ultimately confirmed the potential of fine-tuned LLMs in streamlining various labor-intensive processes of conducting literature reviews. Given the potential of this approach and its applicability across all research domains, this foundational study also advocated for updating PRISMA reporting guidelines to incorporate AI-driven processes, ensuring methodological transparency and reliability in future SLRs. This study broadens the appeal of AI-enhanced tools across various academic and research fields, setting a new standard for conducting comprehensive and accurate literature reviews with more efficiency in the face of ever-increasing volumes of academic studies.
Authors: Tianyu Chen, Yiming Zhang, Guoxin Yu, Dapeng Zhang, Li Zeng, Qing He, Xiang Ao
Abstract: In this paper, we extend financial sentiment analysis~(FSA) to event-level since events usually serve as the subject of the sentiment in financial text. Though extracting events from the financial text may be conducive to accurate sentiment predictions, it has specialized challenges due to the lengthy and discontinuity of events in a financial text. To this end, we reconceptualize the event extraction as a classification task by designing a categorization comprising coarse-grained and fine-grained event categories. Under this setting, we formulate the \textbf{E}vent-Level \textbf{F}inancial \textbf{S}entiment \textbf{A}nalysis~(\textbf{EFSA} for short) task that outputs quintuples consisting of (company, industry, coarse-grained event, fine-grained event, sentiment) from financial text. A large-scale Chinese dataset containing $12,160$ news articles and $13,725$ quintuples is publicized as a brand new testbed for our task. A four-hop Chain-of-Thought LLM-based approach is devised for this task. Systematically investigations are conducted on our dataset, and the empirical results demonstrate the benchmarking scores of existing methods and our proposed method can reach the current state-of-the-art. Our dataset and framework implementation are available at https://anonymous.4open.science/r/EFSA-645E
Authors: Lucas Jos\'e Gon\c{c}alves Freitas, Tha\'is Rodrigues, Guilherme Rodrigues, Pamella Edokawa, Ariane Farias
Abstract: Data analysis and machine learning are of preeminent importance in the legal domain, especially in tasks like clustering and text classification. In this study, we harnessed the power of natural language processing tools to enhance datasets meticulously curated by experts. This process significantly improved the classification workflow for legal texts using machine learning techniques. We considered the Sustainable Development Goals (SDGs) data from the United Nations 2030 Agenda as a practical case study. Data augmentation clustering-based strategy led to remarkable enhancements in the accuracy and sensitivity metrics of classification models. For certain SDGs within the 2030 Agenda, we observed performance gains of over 15%. In some cases, the example base expanded by a noteworthy factor of 5. When dealing with unclassified legal texts, data augmentation strategies centered around clustering prove to be highly effective. They provide a valuable means to expand the existing knowledge base without the need for labor-intensive manual classification efforts.
Authors: Gian Alexandre Michaelsen, Renato P. dos Santos
Abstract: Background: The integration of artificial intelligence (AI) into daily life, particularly through chatbots utilizing natural language processing (NLP), presents both revolutionary potential and unique challenges. This intended to investigate how different input forms impact ChatGPT, a leading language model by OpenAI, performance in understanding and executing complex, multi-intention tasks. Design: Employing a case study methodology supplemented by discourse analysis, the research analyzes ChatGPT's responses to inputs varying from natural language to pseudo-code engineering. The study specifically examines the model's proficiency across four categories: understanding of intentions, interpretability, completeness, and creativity. Setting and Participants: As a theoretical exploration of AI interaction, this study focuses on the analysis of structured and unstructured inputs processed by ChatGPT, without direct human participants. Data collection and analysis: The research utilizes synthetic case scenarios, including the organization of a "weekly meal plan" and a "shopping list," to assess ChatGPT's response to prompts in both natural language and pseudo-code engineering. The analysis is grounded in the identification of patterns, contradictions, and unique response elements across different input formats. Results: Findings reveal that pseudo-code engineering inputs significantly enhance the clarity and determinism of ChatGPT's responses, reducing ambiguity inherent in natural language. Enhanced natural language, structured through prompt engineering techniques, similarly improves the model's interpretability and creativity. Conclusions: The study underscores the potential of pseudo-code engineering in refining human-AI interaction and achieving more deterministic, concise, and direct outcomes, advocating for its broader application across disciplines requiring precise AI responses.
Authors: Bhavith Chandra Challagundla, Chakradhar Peddavenkatagari
Abstract: Automatic text summarization (TS) plays a pivotal role in condensing large volumes of information into concise, coherent summaries, facilitating efficient information retrieval and comprehension. This paper presents a novel framework for abstractive TS of single documents, which integrates three dominant aspects: structural, semantic, and neural-based approaches. The proposed framework merges machine learning and knowledge-based techniques to achieve a unified methodology. The framework consists of three main phases: pre-processing, machine learning, and post-processing. In the pre-processing phase, a knowledge-based Word Sense Disambiguation (WSD) technique is employed to generalize ambiguous words, enhancing content generalization. Semantic content generalization is then performed to address out-of-vocabulary (OOV) or rare words, ensuring comprehensive coverage of the input document. Subsequently, the generalized text is transformed into a continuous vector space using neural language processing techniques. A deep sequence-to-sequence (seq2seq) model with an attention mechanism is employed to predict a generalized summary based on the vector representation. In the post-processing phase, heuristic algorithms and text similarity metrics are utilized to refine the generated summary further. Concepts from the generalized summary are matched with specific entities, enhancing coherence and readability. Experimental evaluations conducted on prominent datasets, including Gigaword, Duc 2004, and CNN/DailyMail, demonstrate the effectiveness of the proposed framework. Results indicate significant improvements in handling rare and OOV words, outperforming existing state-of-the-art deep learning techniques. The proposed framework presents a comprehensive and unified approach towards abstractive TS, combining the strengths of structure, semantics, and neural-based methodologies.
Authors: Chanwoo Choi
Abstract: This work demonstrates two Privacy Policy (PP) summarisation models based on two different clustering algorithms: K-means clustering and Pre-determined Centroid (PDC) clustering. K-means is decided to be used for the first model after an extensive evaluation of ten commonly used clustering algorithms. The summariser model based on the PDC-clustering algorithm summarises PP documents by segregating individual sentences by Euclidean distance from each sentence to the pre-defined cluster centres. The cluster centres are defined according to General Data Protection Regulation (GDPR)'s 14 essential topics that must be included in any privacy notices. The PDC model outperformed the K-means model for two evaluation methods, Sum of Squared Distance (SSD) and ROUGE by some margin (27% and 24% respectively). This result contrasts the K-means model's better performance in the general clustering of sentence vectors before running the task-specific evaluation. This indicates the effectiveness of operating task-specific fine-tuning measures on unsupervised machine-learning models. The summarisation mechanisms implemented in this paper demonstrates an idea of how to efficiently extract essential sentences that should be included in any PP documents. The summariser models could be further developed to an application that tests the GDPR-compliance (or any data privacy legislation) of PP documents.
Authors: Dmitriy Bespalov, Sourav Bhabesh, Yi Xiang, Liutong Zhou, Yanjun Qi
Abstract: Recent NLP literature pays little attention to the robustness of toxicity language predictors, while these systems are most likely to be used in adversarial contexts. This paper presents a novel adversarial attack, \texttt{ToxicTrap}, introducing small word-level perturbations to fool SOTA text classifiers to predict toxic text samples as benign. ToxicTrap exploits greedy based search strategies to enable fast and effective generation of toxic adversarial examples. Two novel goal function designs allow ToxicTrap to identify weaknesses in both multiclass and multilabel toxic language detectors. Our empirical results show that SOTA toxicity text classifiers are indeed vulnerable to the proposed attacks, attaining over 98\% attack success rates in multilabel cases. We also show how a vanilla adversarial training and its improved version can help increase robustness of a toxicity detector even against unseen attacks.
Authors: Feihu Jiang, Chuan Qin, Kaichun Yao, Chuyu Fang, Fuzhen Zhuang, Hengshu Zhu, Hui Xiong
Abstract: Efficient knowledge management plays a pivotal role in augmenting both the operational efficiency and the innovative capacity of businesses and organizations. By indexing knowledge through vectorization, a variety of knowledge retrieval methods have emerged, significantly enhancing the efficacy of knowledge management systems. Recently, the rapid advancements in generative natural language processing technologies paved the way for generating precise and coherent answers after retrieving relevant documents tailored to user queries. However, for enterprise knowledge bases, assembling extensive training data from scratch for knowledge retrieval and generation is a formidable challenge due to the privacy and security policies of private data, frequently entailing substantial costs. To address the challenge above, in this paper, we propose EKRG, a novel Retrieval-Generation framework based on large language models (LLMs), expertly designed to enable question-answering for Enterprise Knowledge bases with limited annotation costs. Specifically, for the retrieval process, we first introduce an instruction-tuning method using an LLM to generate sufficient document-question pairs for training a knowledge retriever. This method, through carefully designed instructions, efficiently generates diverse questions for enterprise knowledge bases, encompassing both fact-oriented and solution-oriented knowledge. Additionally, we develop a relevance-aware teacher-student learning strategy to further enhance the efficiency of the training process. For the generation process, we propose a novel chain of thought (CoT) based fine-tuning method to empower the LLM-based generator to adeptly respond to user questions using retrieved documents. Finally, extensive experiments on real-world datasets have demonstrated the effectiveness of our proposed framework.
Authors: Jie Ou, Yueming Chen, Wenhong Tian
Abstract: While Large Language Models (LLMs) have shown remarkable abilities, they are hindered by significant resource consumption and considerable latency due to autoregressive processing. In this study, we introduce Adaptive N-gram Parallel Decoding (ANPD), an innovative and lossless approach that accelerates inference by allowing the simultaneous generation of multiple tokens. ANPD incorporates a two-stage approach: it begins with a rapid drafting phase that employs an N-gram module, which adapts based on the current interactive context, followed by a verification phase, during which the original LLM assesses and confirms the proposed tokens. Consequently, ANPD preserves the integrity of the LLM's original output while enhancing processing speed. We further leverage a multi-level architecture for the N-gram module to enhance the precision of the initial draft, consequently reducing inference latency. ANPD eliminates the need for retraining or extra GPU memory, making it an efficient and plug-and-play enhancement. In our experiments, models such as LLaMA and its fine-tuned variants have shown speed improvements up to 3.67x, validating the effectiveness of our proposed ANPD.
Authors: Ahmed Agiza, Mohamed Mostagir, Sherief Reda
Abstract: In an era where language models are increasingly integrated into decision-making and communication, understanding the biases within Large Language Models (LLMs) becomes imperative, especially when these models are applied in the economic and political domains. This work investigates the impact of fine-tuning and data selection on economic and political biases in LLM. We explore the methodological aspects of biasing LLMs towards specific ideologies, mindful of the biases that arise from their extensive training on diverse datasets. Our approach, distinct from earlier efforts that either focus on smaller models or entail resource-intensive pre-training, employs Parameter-Efficient Fine-Tuning (PEFT) techniques. These techniques allow for the alignment of LLMs with targeted ideologies by modifying a small subset of parameters. We introduce a systematic method for dataset selection, annotation, and instruction tuning, and we assess its effectiveness through both quantitative and qualitative evaluations. Our work analyzes the potential of embedding specific biases into LLMs and contributes to the dialogue on the ethical application of AI, highlighting the importance of deploying AI in a manner that aligns with societal values.
Authors: Seyed Mahed Mousavi, Simone Alghisi, Giuseppe Riccardi
Abstract: We study the appropriateness of Large Language Models (LLMs) as knowledge repositories. We focus on the challenge of maintaining LLMs' factual knowledge up-to-date over time. Motivated by the lack of studies on identifying outdated knowledge within LLMs, we design and develop a dynamic benchmark with up-to-date ground truth answers for each target factual question. We evaluate eighteen open-source and closed-source state-of-the-art LLMs on time-sensitive knowledge retrieved in real-time from Wikidata. We select time-sensitive domain facts in politics, sports, and organizations, and estimate the recency of the information learned by the model during pre-training\fine-tuning. In the second contribution, we evaluate the effectiveness of knowledge editing methods for aligning LLMs with up-to-date factual knowledge and compare their performance with Retrieval Augmented Generation. The dynamic benchmark is designed to be used as-is to assess LLMs's up-to-dateness, as well as to be extended to other domains by sharing the code, the dataset, as well as evaluation and visualization scripts.
Authors: Avinash Anand, Janak Kapuriya, Apoorv Singh, Jay Saraf, Naman Lal, Astha Verma, Rushali Gupta, Rajiv Shah
Abstract: While Large Language Models (LLMs) can achieve human-level performance in various tasks, they continue to face challenges when it comes to effectively tackling multi-step physics reasoning tasks. To identify the shortcomings of existing models and facilitate further research in this area, we curated a novel dataset, MM-PhyQA, which comprises well-constructed, high schoollevel multimodal physics problems. By evaluating the performance of contemporary LLMs that are publicly available, both with and without the incorporation of multimodal elements in these problems, we aim to shed light on their capabilities. For generating answers for questions consisting of multimodal input (in this case, images and text) we employed Zero-shot prediction using GPT-4 and utilized LLaVA (LLaVA and LLaVA-1.5), the latter of which were fine-tuned on our dataset. For evaluating the performance of LLMs consisting solely of textual input, we tested the performance of the base and fine-tuned versions of the Mistral-7B and LLaMA2-7b models. We also showcased the performance of the novel Multi-Image Chain-of-Thought (MI-CoT) Prompting technique, which when used to train LLaVA-1.5 13b yielded the best results when tested on our dataset, with superior scores in most metrics and the highest accuracy of 71.65% on the test set.
Authors: Agasthya Gangavarapu
Abstract: Addressing the imminent shortfall of 10 million health workers by 2030, predominantly in Low- and Middle-Income Countries (LMICs), this paper introduces an innovative approach that harnesses the power of Large Language Models (LLMs) integrated with machine translation models. This solution is engineered to meet the unique needs of Community Health Workers (CHWs), overcoming language barriers, cultural sensitivities, and the limited availability of medical dialog datasets. I have crafted a model that not only boasts superior translation capabilities but also undergoes rigorous fine-tuning on open-source datasets to ensure medical accuracy and is equipped with comprehensive safety features to counteract the risks of misinformation. Featuring a modular design, this approach is specifically structured for swift adaptation across various linguistic and cultural contexts, utilizing open-source components to significantly reduce healthcare operational costs. This strategic innovation markedly improves the accessibility and quality of healthcare services by providing CHWs with contextually appropriate medical knowledge and diagnostic tools. This paper highlights the transformative impact of this context-aware LLM, underscoring its crucial role in addressing the global healthcare workforce deficit and propelling forward healthcare outcomes in LMICs.
Authors: Siyang Liu, Trish Maturi, Siqi Shen, Rada Mihalcea
Abstract: In this paper, we explore the alignment of values in Large Language Models (LLMs) with specific age groups, leveraging data from the World Value Survey across thirteen categories. Through a diverse set of prompts tailored to ensure response robustness, we find a general inclination of LLM values towards younger demographics. Additionally, we explore the impact of incorporating age identity information in prompts and observe challenges in mitigating value discrepancies with different age cohorts. Our findings highlight the age bias in LLMs and provide insights for future work.
Authors: Matthew DeLorenzo, Vasudev Gohil, Jeyavijayan Rajendran
Abstract: Large Language Models (LLMs) have proved effective and efficient in generating code, leading to their utilization within the hardware design process. Prior works evaluating LLMs' abilities for register transfer level code generation solely focus on functional correctness. However, the creativity associated with these LLMs, or the ability to generate novel and unique solutions, is a metric not as well understood, in part due to the challenge of quantifying this quality. To address this research gap, we present CreativeEval, a framework for evaluating the creativity of LLMs within the context of generating hardware designs. We quantify four creative sub-components, fluency, flexibility, originality, and elaboration, through various prompting and post-processing techniques. We then evaluate multiple popular LLMs (including GPT models, CodeLlama, and VeriGen) upon this creativity metric, with results indicating GPT-3.5 as the most creative model in generating hardware designs.
Authors: R. Michael Alvarez, Jacob Morrier
Abstract: This paper presents a new approach to evaluating the quality of answers in political question-and-answer sessions. We propose to measure an answer's quality based on the degree to which it allows us to infer the initial question accurately. This conception of answer quality inherently reflects their relevance to initial questions. Drawing parallels with semantic search, we argue that this measurement approach can be operationalized by fine-tuning a large language model on the observed corpus of questions and answers without additional labeled data. We showcase our measurement approach within the context of the Question Period in the Canadian House of Commons. Our approach yields valuable insights into the correlates of the quality of answers in the Question Period. We find that answer quality varies significantly based on the party affiliation of the members of Parliament asking the questions and uncover a meaningful correlation between answer quality and the topics of the questions.
Authors: Yewei Song, Cedric Lothritz, Daniel Tang, Tegawend\'e F. Bissyand\'e, Jacques Klein
Abstract: This paper revisits recent code similarity evaluation metrics, particularly focusing on the application of Abstract Syntax Tree (AST) editing distance in diverse programming languages. In particular, we explore the usefulness of these metrics and compare them to traditional sequence similarity metrics. Our experiments showcase the effectiveness of AST editing distance in capturing intricate code structures, revealing a high correlation with established metrics. Furthermore, we explore the strengths and weaknesses of AST editing distance and prompt-based GPT similarity scores in comparison to BLEU score, execution match, and Jaccard Similarity. We propose, optimize, and publish an adaptable metric that demonstrates effectiveness across all tested languages, representing an enhanced version of Tree Similarity of Edit Distance (TSED).
Authors: Ji-Ung Lee, Marc E. Pfetsch, Iryna Gurevych
Abstract: This work proposes a novel method to generate C-Tests; a deviated form of cloze tests (a gap filling exercise) where only the last part of a word is turned into a gap. In contrast to previous works that only consider varying the gap size or gap placement to achieve locally optimal solutions, we propose a mixed-integer programming (MIP) approach. This allows us to consider gap size and placement simultaneously, achieving globally optimal solutions, and to directly integrate state-of-the-art models for gap difficulty prediction into the optimization problem. A user study with 40 participants across four C-Test generation strategies (including GPT-4) shows that our approach (MIP) significantly outperforms two of the baseline strategies (based on gap placement and GPT-4); and performs on-par with the third (based on gap size). Our analysis shows that GPT-4 still struggles to fulfill explicit constraints during generation and that MIP produces C-Tests that correlate best with the perceived difficulty. We publish our code, model, and collected data consisting of 32 English C-Tests with 20 gaps each (totaling 3,200 individual gap responses) under an open source license.
Authors: Zezheng Li, Kingston Yip
Abstract: This study introduces a novel BERT-LSH model that incorporates Locality Sensitive Hashing (LSH) to approximate the attention mechanism in the BERT architecture. We examine the computational efficiency and performance of this model compared to a standard baseline BERT model. Our findings reveal that BERT-LSH significantly reduces computational demand for the self-attention layer while unexpectedly outperforming the baseline model in pretraining and fine-tuning tasks. These results suggest that the LSH-based attention mechanism not only offers computational advantages but also may enhance the model's ability to generalize from its training data. For more information, visit our GitHub repository: https://github.com/leo4life2/algoml-final
Authors: Mukul Gagrani, Raghavv Goel, Wonseok Jeon, Junyoung Park, Mingu Lee, Christopher Lott
Abstract: Inference with Multimodal Large Language Models (MLLMs) is slow due to their large-language-model backbone which suffers from memory bandwidth bottleneck and generates tokens auto-regressively. In this paper, we explore the application of speculative decoding to enhance the inference efficiency of MLLMs, specifically the LLaVA 7B model. We show that a language-only model can serve as a good draft model for speculative decoding with LLaVA 7B, bypassing the need for image tokens and their associated processing components from the draft model. Our experiments across three different tasks show that speculative decoding can achieve a memory-bound speedup of up to 2.37$\times$ using a 115M parameter language model that we trained from scratch. Additionally, we introduce a compact LLaVA draft model incorporating an image adapter, which shows marginal performance gains in image captioning while maintaining comparable results in other tasks.
Authors: Daniel Machlab, Rick Battle
Abstract: The proliferation of Large Language Models (LLMs) highlights the critical importance of conducting thorough evaluations to discern their comparative advantages, limitations, and optimal use cases. Particularly important is assessing their capacity to accurately retrieve information included in a given prompt. A model's ability to do this significantly influences how effectively it can utilize contextual details, thus impacting its practical efficacy and dependability in real-world applications. Our research analyzes the in-context recall performance of various LLMs using the needle-in-a-haystack method. In this approach, a factoid (the "needle") is embedded within a block of filler text (the "haystack"), which the model is asked to retrieve. We assess the recall performance of each model across various haystack lengths and with varying needle placements to identify performance patterns. This study demonstrates that an LLM's recall capability is not only contingent upon the prompt's content but also may be compromised by biases in its training data. Conversely, adjustments to model architecture, training strategy, or fine-tuning can improve performance. Our analysis provides insight into LLM behavior, offering direction for the development of more effective applications of LLMs.
Authors: Yue Zhou, Barbara Di Eugenio, Brian Ziebart, Lisa Sharp, Bing Liu, Ben Gerber, Nikolaos Agadakos, Shweta Yadav
Abstract: Health coaching helps patients identify and accomplish lifestyle-related goals, effectively improving the control of chronic diseases and mitigating mental health conditions. However, health coaching is cost-prohibitive due to its highly personalized and labor-intensive nature. In this paper, we propose to build a dialogue system that converses with the patients, helps them create and accomplish specific goals, and can address their emotions with empathy. However, building such a system is challenging since real-world health coaching datasets are limited and empathy is subtle. Thus, we propose a modularized health coaching dialogue system with simplified NLU and NLG frameworks combined with mechanism-conditioned empathetic response generation. Through automatic and human evaluation, we show that our system generates more empathetic, fluent, and coherent responses and outperforms the state-of-the-art in NLU tasks while requiring less annotation. We view our approach as a key step towards building automated and more accessible health coaching systems.
Authors: Wei Zou, Ziyuan Zhuang, Shujian Huang, Jia Liu, Jiajun Chen
Abstract: Paraphrase generation aims to produce high-quality and diverse utterances of a given text. Though state-of-the-art generation via the diffusion model reconciles generation quality and diversity, textual diffusion suffers from a truncation issue that hinders efficiency and quality control. In this work, we propose \textit{L}atent \textit{D}iffusion \textit{P}araphraser~(LDP), a novel paraphrase generation by modeling a controllable diffusion process given a learned latent space. LDP achieves superior generation efficiency compared to its diffusion counterparts. It facilitates only input segments to enforce paraphrase semantics, which further improves the results without external features. Experiments show that LDP achieves improved and diverse paraphrase generation compared to baselines. Further analysis shows that our method is also helpful to other similar text generations and domain adaptations. Our code and data are available at https://github.com/NIL-zhuang/ld4pg.
Authors: Abhijnan Nath, Huma Jamil, Shafiuddin Rehan Ahmed, George Baker, Rahul Ghosh, James H. Martin, Nathaniel Blanchard, Nikhil Krishnaswamy
Abstract: Event coreference resolution (ECR) is the task of determining whether distinct mentions of events within a multi-document corpus are actually linked to the same underlying occurrence. Images of the events can help facilitate resolution when language is ambiguous. Here, we propose a multimodal cross-document event coreference resolution method that integrates visual and textual cues with a simple linear map between vision and language models. As existing ECR benchmark datasets rarely provide images for all event mentions, we augment the popular ECB+ dataset with event-centric images scraped from the internet and generated using image diffusion models. We establish three methods that incorporate images and text for coreference: 1) a standard fused model with finetuning, 2) a novel linear mapping method without finetuning and 3) an ensembling approach based on splitting mention pairs by semantic and discourse-level difficulty. We evaluate on 2 datasets: the augmented ECB+, and AIDA Phase 1. Our ensemble systems using cross-modal linear mapping establish an upper limit (91.9 CoNLL F1) on ECB+ ECR performance given the preprocessing assumptions used, and establish a novel baseline on AIDA Phase 1. Our results demonstrate the utility of multimodal information in ECR for certain challenging coreference problems, and highlight a need for more multimodal resources in the coreference resolution space.
Authors: Tom\'a\v{s} Sourada, Jana Strakov\'a, Rudolf Rosa
Abstract: We focus on morphological inflection in out-of-vocabulary (OOV) conditions, an under-researched subtask in which state-of-the-art systems usually are less effective. We developed three systems: a retrograde model and two sequence-to-sequence (seq2seq) models based on LSTM and Transformer. For testing in OOV conditions, we automatically extracted a large dataset of nouns in the morphologically rich Czech language, with lemma-disjoint data splits, and we further manually annotated a real-world OOV dataset of neologisms. In the standard OOV conditions, Transformer achieves the best results, with increasing performance in ensemble with LSTM, the retrograde model and SIGMORPHON baselines. On the real-world OOV dataset of neologisms, the retrograde model outperforms all neural models. Finally, our seq2seq models achieve state-of-the-art results in 9 out of 16 languages from SIGMORPHON 2022 shared task data in the OOV evaluation (feature overlap) in the large data condition. We release the Czech OOV Inflection Dataset for rigorous evaluation in OOV conditions. Further, we release the inflection system with the seq2seq models as a ready-to-use Python library.
Authors: Shun Zhang, Chaoran Yan, Jian Yang, Changyu Ren, Jiaqi Bai, Tongliang Li, Zhoujun Li
Abstract: New Intent Discovery (NID) strives to identify known and reasonably deduce novel intent groups in the open-world scenario. But current methods face issues with inaccurate pseudo-labels and poor representation learning, creating a negative feedback loop that degrades overall model performance, including accuracy and the adjusted rand index. To address the aforementioned challenges, we propose a Robust New Intent Discovery (RoNID) framework optimized by an EM-style method, which focuses on constructing reliable pseudo-labels and obtaining cluster-friendly discriminative representations. RoNID comprises two main modules: reliable pseudo-label generation module and cluster-friendly representation learning module. Specifically, the pseudo-label generation module assigns reliable synthetic labels by solving an optimal transport problem in the E-step, which effectively provides high-quality supervised signals for the input of the cluster-friendly representation learning module. To learn cluster-friendly representation with strong intra-cluster compactness and large inter-cluster separation, the representation learning module combines intra-cluster and inter-cluster contrastive learning in the M-step to feed more discriminative features into the generation module. RoNID can be performed iteratively to ultimately yield a robust model with reliable pseudo-labels and cluster-friendly representations. Experimental results on multiple benchmarks demonstrate our method brings substantial improvements over previous state-of-the-art methods by a large margin of +1~+4 points.
Authors: Ryan Cotterell, Thomas M\"uller, Alexander Fraser, Hinrich Sch\"utze
Abstract: We present labeled morphological segmentation, an alternative view of morphological processing that unifies several tasks. From an annotation standpoint, we additionally introduce a new hierarchy of morphotactic tagsets. Finally, we develop \modelname, a discriminative morphological segmentation system that, contrary to previous work, explicitly models morphotactics. We show that \textsc{chipmunk} yields improved performance on three tasks for all six languages: (i) morphological segmentation, (ii) stemming and (iii) morphological tag classification. On morphological segmentation, our method shows absolute improvements of 2--6 points $F_1$ over the baseline.
Authors: Hayato Tsukagoshi, Tsutomu Hirao, Makoto Morishita, Katsuki Chousa, Ryohei Sasano, Koichi Takeda
Abstract: The task of Split and Rephrase, which splits a complex sentence into multiple simple sentences with the same meaning, improves readability and enhances the performance of downstream tasks in natural language processing (NLP). However, while Split and Rephrase can be improved using a text-to-text generation approach that applies encoder-decoder models fine-tuned with a large-scale dataset, it still suffers from hallucinations and under-splitting. To address these issues, this paper presents a simple and strong data refinement approach. Here, we create WikiSplit++ by removing instances in WikiSplit where complex sentences do not entail at least one of the simpler sentences and reversing the order of reference simple sentences. Experimental results show that training with WikiSplit++ leads to better performance than training with WikiSplit, even with fewer training instances. In particular, our approach yields significant gains in the number of splits and the entailment ratio, a proxy for measuring hallucinations.
Authors: Yusheng Liao, Shuyang Jiang, Yu Wang, Yanfeng Wang
Abstract: Large language models like ChatGPT have shown substantial progress in natural language understanding and generation, proving valuable across various disciplines, including the medical field. Despite advancements, challenges persist due to the complexity and diversity inherent in medical tasks which often require multi-task learning capabilities. Previous approaches, although beneficial, fall short in real-world applications because they necessitate task-specific annotations at inference time, limiting broader generalization. This paper introduces MING-MOE, a novel Mixture-of-Expert~(MOE)-based medical large language model designed to manage diverse and complex medical tasks without requiring task-specific annotations, thus enhancing its usability across extensive datasets. MING-MOE employs a Mixture of Low-Rank Adaptation (MoLoRA) technique, allowing for efficient parameter usage by maintaining base model parameters static while adapting through a minimal set of trainable parameters. We demonstrate that MING-MOE achieves state-of-the-art (SOTA) performance on over 20 medical tasks, illustrating a significant improvement over existing models. This approach not only extends the capabilities of medical language models but also improves inference efficiency.
Authors: Jia Gu, Liang Pang, Huawei Shen, Xueqi Cheng
Abstract: With the rapid advancement of large language models (LLMs) and their remarkable capabilities in handling complex language tasks, an increasing number of studies are employing LLMs as agents to emulate the sequential decision-making processes of humans often represented as Markov decision-making processes (MDPs). The actions within this decision-making framework adhere to specific probability distributions and require iterative sampling. This arouses our curiosity regarding the capacity of LLM agents to comprehend probability distributions, thereby guiding the agent's behavioral decision-making through probabilistic sampling and generating behavioral sequences. To answer the above question, we divide the problem into two main aspects: simulation where the exact probability distribution is known, and generation of sequences where the probability distribution is ambiguous. In the first case, the agent is required to give the type and parameters of the probability distribution through the problem description, and then give the sampling sequence. However, our analysis shows that LLM agents perform poorly in this case, but the sampling success rate can be improved through programming tools. Real-world scenarios often entail unknown probability distributions. Thus, in the second case, we ask the agents to change the activity level in online social networks and analyze the frequency of actions. Ultimately, our analysis shows that LLM agents cannot sample probability distributions even using programming tools. Therefore, careful consideration is still required before directly applying LLM agents as agents to simulate human behavior.
Authors: Zita Lifelo, Huansheng Ning, Sahraoui Dhelim
Abstract: Timely identification is essential for the efficient handling of mental health illnesses such as depression. However, the current research fails to adequately address the prediction of mental health conditions from social media data in low-resource African languages like Swahili. This study introduces two distinct approaches utilising model-agnostic meta-learning and leveraging large language models (LLMs) to address this gap. Experiments are conducted on three datasets translated to low-resource language and applied to four mental health tasks, which include stress, depression, depression severity and suicidal ideation prediction. we first apply a meta-learning model with self-supervision, which results in improved model initialisation for rapid adaptation and cross-lingual transfer. The results show that our meta-trained model performs significantly better than standard fine-tuning methods, outperforming the baseline fine-tuning in macro F1 score with 18\% and 0.8\% over XLM-R and mBERT. In parallel, we use LLMs' in-context learning capabilities to assess their performance accuracy across the Swahili mental health prediction tasks by analysing different cross-lingual prompting approaches. Our analysis showed that Swahili prompts performed better than cross-lingual prompts but less than English prompts. Our findings show that in-context learning can be achieved through cross-lingual transfer through carefully crafted prompt templates with examples and instructions.
Authors: Sharvi Endait, Srushti Sonavane, Ridhima Sinare, Pritika Rohera, Advait Naik, Dipali Kadam
Abstract: The explosive growth of online content demands robust Natural Language Processing (NLP) techniques that can capture nuanced meanings and cultural context across diverse languages. Semantic Textual Relatedness (STR) goes beyond superficial word overlap, considering linguistic elements and non-linguistic factors like topic, sentiment, and perspective. Despite its pivotal role, prior NLP research has predominantly focused on English, limiting its applicability across languages. Addressing this gap, our paper dives into capturing deeper connections between sentences beyond simple word overlap. Going beyond English-centric NLP research, we explore STR in Marathi, Hindi, Spanish, and English, unlocking the potential for information retrieval, machine translation, and more. Leveraging the SemEval-2024 shared task, we explore various language models across three learning paradigms: supervised, unsupervised, and cross-lingual. Our comprehensive methodology gains promising results, demonstrating the effectiveness of our approach. This work aims to not only showcase our achievements but also inspire further research in multilingual STR, particularly for low-resourced languages.
Authors: Zukang Yang, Zixuan Zhu
Abstract: In the field of Question Answering (QA), unifying large language models (LLMs) with external databases has shown great success. However, these methods often fall short in providing the advanced reasoning needed for complex QA tasks. To address these issues, we improve over a novel approach called Knowledge Graph Prompting (KGP), which combines knowledge graphs with a LLM-based agent to improve reasoning and search accuracy. Nevertheless, the original KGP framework necessitates costly fine-tuning with large datasets yet still suffers from LLM hallucination. Therefore, we propose a reasoning-infused LLM agent to enhance this framework. This agent mimics human curiosity to ask follow-up questions to more efficiently navigate the search. This simple modification significantly boosts the LLM performance in QA tasks without the high costs and latency associated with the initial KGP framework. Our ultimate goal is to further develop this approach, leading to more accurate, faster, and cost-effective solutions in the QA domain.
Authors: Ruixin Yang, Dheeraj Rajagopa, Shirley Anugrah Hayati, Bin Hu, Dongyeop Kang
Abstract: Uncertainty estimation is a significant issue for current large language models (LLMs) that are generally poorly calibrated and over-confident, especially with reinforcement learning from human feedback (RLHF). Unlike humans, whose decisions and confidences not only stem from intrinsic beliefs but can also be adjusted through daily observations, existing calibration methods for LLMs focus on estimating or eliciting individual confidence without taking full advantage of the "Collective Wisdom": the interaction among multiple LLMs that can collectively improve both accuracy and calibration. In this work, we propose Collaborative Calibration, a post-hoc training-free calibration strategy that leverages the collaborative and expressive capabilities of multiple tool-augmented LLM agents in a simulated group deliberation process. We demonstrate the effectiveness of Collaborative Calibration on generative QA tasks across various domains, showing its potential in harnessing the rationalization of collectively calibrated confidence assessments and improving the reliability of model predictions.
Authors: Yanhong Li, Chenghao Yang, Allyson Ettinger
Abstract: Recent studies suggest that self-reflective prompting can significantly enhance the reasoning capabilities of Large Language Models (LLMs). However, the use of external feedback as a stop criterion raises doubts about the true extent of LLMs' ability to emulate human-like self-reflection. In this paper, we set out to clarify these capabilities under a more stringent evaluation setting in which we disallow any kind of external feedback. Our findings under this setting show a split: while self-reflection enhances performance in TruthfulQA, it adversely affects results in HotpotQA. We conduct follow-up analyses to clarify the contributing factors in these patterns, and find that the influence of self-reflection is impacted both by reliability of accuracy in models' initial responses, and by overall question difficulty: specifically, self-reflection shows the most benefit when models are less likely to be correct initially, and when overall question difficulty is higher. We also find that self-reflection reduces tendency toward majority voting. Based on our findings, we propose guidelines for decisions on when to implement self-reflection. We release the codebase for reproducing our experiments at https://github.com/yanhong-lbh/LLM-SelfReflection-Eval.
URLs: https://github.com/yanhong-lbh/LLM-SelfReflection-Eval.
Authors: Taojun Hu, Xiao-Hua Zhou
Abstract: Natural Language Processing (NLP) is witnessing a remarkable breakthrough driven by the success of Large Language Models (LLMs). LLMs have gained significant attention across academia and industry for their versatile applications in text generation, question answering, and text summarization. As the landscape of NLP evolves with an increasing number of domain-specific LLMs employing diverse techniques and trained on various corpus, evaluating performance of these models becomes paramount. To quantify the performance, it's crucial to have a comprehensive grasp of existing metrics. Among the evaluation, metrics which quantifying the performance of LLMs play a pivotal role. This paper offers a comprehensive exploration of LLM evaluation from a metrics perspective, providing insights into the selection and interpretation of metrics currently in use. Our main goal is to elucidate their mathematical formulations and statistical interpretations. We shed light on the application of these metrics using recent Biomedical LLMs. Additionally, we offer a succinct comparison of these metrics, aiding researchers in selecting appropriate metrics for diverse tasks. The overarching goal is to furnish researchers with a pragmatic guide for effective LLM evaluation and metric selection, thereby advancing the understanding and application of these large language models.
Authors: Spandan Das, Vinay Samuel, Shahriar Noroozizadeh
Abstract: This paper introduces novel methodologies for the Natural Language Inference for Clinical Trials (NLI4CT) task. We present TLDR (T5-generated clinical-Language summaries for DeBERTa Report Analysis) which incorporates T5-model generated premise summaries for improved entailment and contradiction analysis in clinical NLI tasks. This approach overcomes the challenges posed by small context windows and lengthy premises, leading to a substantial improvement in Macro F1 scores: a 0.184 increase over truncated premises. Our comprehensive experimental evaluation, including detailed error analysis and ablations, confirms the superiority of TLDR in achieving consistency and faithfulness in predictions against semantically altered inputs.
Authors: Artur Kiulian, Anton Polishko, Mykola Khandoga, Oryna Chubych, Jack Connor, Raghav Ravishankar, Adarsh Shirawalmath
Abstract: In the rapidly advancing field of AI and NLP, generative large language models (LLMs) stand at the forefront of innovation, showcasing unparalleled abilities in text understanding and generation. However, the limited representation of low-resource languages like Ukrainian poses a notable challenge, restricting the reach and relevance of this technology. Our paper addresses this by fine-tuning the open-source Gemma and Mistral LLMs with Ukrainian datasets, aiming to improve their linguistic proficiency and benchmarking them against other existing models capable of processing Ukrainian language. This endeavor not only aims to mitigate language bias in technology but also promotes inclusivity in the digital realm. Our transparent and reproducible approach encourages further NLP research and development. Additionally, we present the Ukrainian Knowledge and Instruction Dataset (UKID) to aid future efforts in language model fine-tuning. Our research not only advances the field of NLP but also highlights the importance of linguistic diversity in AI, which is crucial for cultural preservation, education, and expanding AI's global utility. Ultimately, we advocate for a future where technology is inclusive, enabling AI to communicate effectively across all languages, especially those currently underrepresented.
Authors: Guochao Jiang, Ziqin Luo, Yuchen Shi, Dixuan Wang, Jiaqing Liang, Deqing Yang
Abstract: In recent years, the fine-tuned generative models have been proven more powerful than the previous tagging-based or span-based models on named entity recognition (NER) task. It has also been found that the information related to entities, such as entity types, can prompt a model to achieve NER better. However, it is not easy to determine the entity types indeed existing in the given sentence in advance, and inputting too many potential entity types would distract the model inevitably. To exploit entity types' merit on promoting NER task, in this paper we propose a novel NER framework, namely ToNER based on a generative model. In ToNER, a type matching model is proposed at first to identify the entity types most likely to appear in the sentence. Then, we append a multiple binary classification task to fine-tune the generative model's encoder, so as to generate the refined representation of the input sentence. Moreover, we add an auxiliary task for the model to discover the entity types which further fine-tunes the model to output more accurate results. Our extensive experiments on some NER benchmarks verify the effectiveness of our proposed strategies in ToNER that are oriented towards entity types' exploitation.
Authors: Amani Namboori, Shivam Mangale, Andy Rosenbaum, Saleh Soltan
Abstract: The emergence of Large Language Models (LLMs) with capabilities like In-Context Learning (ICL) has ushered in new possibilities for data generation across various domains while minimizing the need for extensive data collection and modeling techniques. Researchers have explored ways to use this generated synthetic data to optimize smaller student models for reduced deployment costs and lower latency in downstream tasks. However, ICL-generated data often suffers from low quality as the task specificity is limited with few examples used in ICL. In this paper, we propose GeMQuAD - a semi-supervised learning approach, extending the WeakDAP framework, applied to a dataset generated through ICL with just one example in the target language using AlexaTM 20B Seq2Seq LLM. Through our approach, we iteratively identify high-quality data to enhance model performance, especially for low-resource multilingual setting in the context of Extractive Question Answering task. Our framework outperforms the machine translation-augmented model by 0.22/1.68 F1/EM (Exact Match) points for Hindi and 0.82/1.37 F1/EM points for Spanish on the MLQA dataset, and it surpasses the performance of model trained on an English-only dataset by 5.05/6.50 F1/EM points for Hindi and 3.81/3.69 points F1/EM for Spanish on the same dataset. Notably, our approach uses a pre-trained LLM for generation with no fine-tuning (FT), utilizing just a single annotated example in ICL to generate data, providing a cost-effective development process.
Authors: Xiao Chen, Sihang Zhou, Ke Liang, Xinwang Liu
Abstract: Chain of thought finetuning aims to endow small student models with reasoning capacity to improve their performance towards a specific task by allowing them to imitate the reasoning procedure of large language models (LLMs) beyond simply predicting the answer to the question. However, the existing methods 1) generate rationale before the answer, making their answer correctness sensitive to the hallucination in the rationale;2) force the student model to repeat the exact LLMs rationale expression word-after-word, which could have the model biased towards learning the expression in rationale but count against the model from understanding the core logic behind it. Therefore, we propose a robust Post-Semantic-Thinking (PST) strategy to generate answers before rationale. Thanks to this answer-first setting, 1) the answering procedure can escape from the adverse effects caused by hallucinations in the rationale; 2) the complex reasoning procedure is tightly bound with the relatively concise answer, making the reasoning for questions easier with the prior information in the answer; 3) the efficiency of the method can also benefit from the setting since users can stop the generation right after answers are outputted when inference is conducted. Furthermore, the PST strategy loose the constraint against the generated rationale to be close to the LLMs gold standard in the hidden semantic space instead of the vocabulary space, thus making the small student model better comprehend the semantic reasoning logic in rationale. Extensive experiments conducted across 12 reasoning tasks demonstrate the effectiveness of PST.
Authors: Yuqi Wang, Zeqiang Wang, Wei Wang, Qi Chen, Kaizhu Huang, Anh Nguyen, Suparna De
Abstract: Safe and reliable natural language inference is critical for extracting insights from clinical trial reports but poses challenges due to biases in large pre-trained language models. This paper presents a novel data augmentation technique to improve model robustness for biomedical natural language inference in clinical trials. By generating synthetic examples through semantic perturbations and domain-specific vocabulary replacement and adding a new task for numerical and quantitative reasoning, we introduce greater diversity and reduce shortcut learning. Our approach, combined with multi-task learning and the DeBERTa architecture, achieved significant performance gains on the NLI4CT 2024 benchmark compared to the original language models. Ablation studies validate the contribution of each augmentation method in improving robustness. Our best-performing model ranked 12th in terms of faithfulness and 8th in terms of consistency, respectively, out of the 32 participants.
Authors: Sophia Maria
Abstract: Large language models have exhibited significant proficiency in languages endowed with extensive linguistic resources, such as English and Chinese. Nevertheless, their effectiveness notably diminishes when applied to languages characterized by limited linguistic resources, particularly within the Southeast Asian linguistic landscape, such as Indonesian. The scarcity of linguistic resources for these languages presents challenges associated with inadequate training, restricted vocabulary coverage, and challenging evaluation processes. In response to these exigencies, we have introduced CompassLLM, a large multilingual model specifically tailored for Southeast Asian languages, with the primary aim of supporting the developmental requirements of Shopee. Our methodology encompasses several key strategies. To progressively enhance multilingual proficiencies, we implemented a multi-stage pre-training strategy integrated with curriculum learning, gradually intensifying the focus on low-resource languages. Concurrently, to better accommodate low-resource human instructions, we curated and generated a repository of high-quality multilingual human instructions, culminating the CompassLLM-SFT model through supervised instruction fine-tuning. Finally, to reinforce the model's alignment with human preference behaviors, we have embraced the principle of Direct Preference Optimization (DPO) to obtain CompassLLM-DPO model. Preliminary evaluation of the CompassLLM model yields promising results, with our model surpassing benchmark models like Vicuna-7b-v1.5, Sealion, Falcon and SeaLLM, across diverse evaluation tasks, as verified through both automated and human-driven assessments. Notably, our model exhibits its superior performance in South-east Asia languages, such as Indonesian language.
Authors: Taehyeon Kim, Ananda Theertha Suresh, Kishore Papineni, Michael Riley, Sanjiv Kumar, Adrian Benton
Abstract: Despite the remarkable strides made by autoregressive language models, their potential is often hampered by the slow inference speeds inherent in sequential token generation. Blockwise parallel decoding (BPD) was proposed by Stern et al. (2018) as a way to improve inference speed of language models. In this paper, we make two contributions to understanding and improving BPD drafts. We first offer an analysis of the token distributions produced by the BPD prediction heads. Secondly, we use this analysis to inform algorithms to improve BPD inference speed by refining the BPD drafts using small n-gram or neural language models. We empirically show that these refined BPD drafts yield a higher average verified prefix length across tasks.
Authors: Kota Tanabe, Masahiro Suzuki, Hiroki Sakaji, Itsuki Noda
Abstract: We construct an instruction dataset for the large language model (LLM) in the Japanese finance domain. Domain adaptation of language models, including LLMs, is receiving more attention as language models become more popular. This study demonstrates the effectiveness of domain adaptation through instruction tuning. To achieve this, we propose an instruction tuning data in Japanese called JaFIn, the Japanese Financial Instruction Dataset. JaFIn is manually constructed based on multiple data sources, including Japanese government websites, which provide extensive financial knowledge. We then utilize JaFIn to apply instruction tuning for several LLMs, demonstrating that our models specialized in finance have better domain adaptability than the original models. The financial-specialized LLMs created were evaluated using a quantitative Japanese financial benchmark and qualitative response comparisons, showing improved performance over the originals.
Authors: Tuan Bui, Oanh Tran, Phuong Nguyen, Bao Ho, Long Nguyen, Thang Bui, Tho Quan
Abstract: In today's rapidly evolving landscape of Artificial Intelligence, large language models (LLMs) have emerged as a vibrant research topic. LLMs find applications in various fields and contribute significantly. Despite their powerful language capabilities, similar to pre-trained language models (PLMs), LLMs still face challenges in remembering events, incorporating new information, and addressing domain-specific issues or hallucinations. To overcome these limitations, researchers have proposed Retrieval-Augmented Generation (RAG) techniques, some others have proposed the integration of LLMs with Knowledge Graphs (KGs) to provide factual context, thereby improving performance and delivering more accurate feedback to user queries. Education plays a crucial role in human development and progress. With the technology transformation, traditional education is being replaced by digital or blended education. Therefore, educational data in the digital environment is increasing day by day. Data in higher education institutions are diverse, comprising various sources such as unstructured/structured text, relational databases, web/app-based API access, etc. Constructing a Knowledge Graph from these cross-data sources is not a simple task. This article proposes a method for automatically constructing a Knowledge Graph from multiple data sources and discusses some initial applications (experimental trials) of KG in conjunction with LLMs for question-answering tasks.
Authors: Dror K. Markus, Effi Levi, Tamir Sheafer, Shaul R. Shenhav
Abstract: Media Storms, dramatic outbursts of attention to a story, are central components of media dynamics and the attention landscape. Despite their significance, there has been little systematic and empirical research on this concept due to issues of measurement and operationalization. We introduce an iterative human-in-the-loop method to identify media storms in a large-scale corpus of news articles. The text is first transformed into signals of dispersion based on several textual characteristics. In each iteration, we apply unsupervised anomaly detection to these signals; each anomaly is then validated by an expert to confirm the presence of a storm, and those results are then used to tune the anomaly detection in the next iteration. We demonstrate the applicability of this method in two scenarios: first, supplementing an initial list of media storms within a specific time frame; and second, detecting media storms in new time periods. We make available a media storm dataset compiled using both scenarios. Both the method and dataset offer the basis for comprehensive empirical research into the concept of media storms, including characterizing them and predicting their outbursts and durations, in mainstream media or social media platforms.
Authors: Carlos Carrasco-Farre
Abstract: Large Language Models (LLMs) are already as persuasive as humans. However, we know very little about why. This paper investigates the persuasion strategies of LLMs, comparing them with human-generated arguments. Using a dataset of 1,251 participants in an experiment, we analyze the persuaion strategies of LLM-generated and human-generated arguments using measures of cognitive effort (lexical and grammatical complexity) and moral-emotional language (sentiment and moral analysis). The study reveals that LLMs produce arguments that require higher cognitive effort, exhibiting more complex grammatical and lexical structures than human counterparts. Additionally, LLMs demonstrate a significant propensity to engage more deeply with moral language, utilizing both positive and negative moral foundations more frequently than humans. In contrast with previous research, no significant difference was found in the emotional content produced by LLMs and humans. These findings contribute to the discourse on AI and persuasion, highlighting the dual potential of LLMs to both enhance and undermine informational integrity through communication strategies for digital persuasion.
Authors: Tian Jin, Wanzin Yazar, Zifei Xu, Sayeh Sharify, Xin Wang
Abstract: Large language models (LLMs) can solve challenging tasks. However, their inference computation on modern GPUs is highly inefficient due to the increasing number of tokens they must attend to as they generate new ones. To address this inefficiency, we capitalize on LLMs' problem-solving capabilities to optimize their own inference-time efficiency. We demonstrate with two specific tasks: (a) evaluating complex arithmetic expressions and (b) summarizing news articles. For both tasks, we create custom datasets to fine-tune an LLM. The goal of fine-tuning is twofold: first, to make the LLM learn to solve the evaluation or summarization task, and second, to train it to identify the minimal attention spans required for each step of the task. As a result, the fine-tuned model is able to convert these self-identified minimal attention spans into sparse attention masks on-the-fly during inference. We develop a custom CUDA kernel to take advantage of the reduced context to attend to. We demonstrate that using this custom CUDA kernel improves the throughput of LLM inference by 28%. Our work presents an end-to-end demonstration showing that training LLMs to self-select their attention spans speeds up autoregressive inference in solving real-world tasks.
Authors: Souvik Das, Lifeng Jin, Linfeng Song, Haitao Mi, Baolin Peng, Dong Yu
Abstract: Large language models (LLMs) exhibit impressive natural language capabilities but suffer from hallucination -- generating content ungrounded in the realities of training data. Recent work has focused on decoding techniques to improve factuality during inference by leveraging LLMs' hierarchical representation of factual knowledge, manipulating the predicted distributions at inference time. Current state-of-the-art approaches refine decoding by contrasting early-exit distributions from a lower layer with the final layer to exploit information related to factuality within the model forward procedure. However, such methods often assume the final layer is the most reliable and the lower layer selection process depends on it. In this work, we first propose extrapolation of critical token probabilities beyond the last layer for more accurate contrasting. We additionally employ layer-wise entropy-guided lower layer selection, decoupling the selection process from the final layer. Experiments demonstrate strong performance - surpassing state-of-the-art on multiple different datasets by large margins. Analyses show different kinds of prompts respond to different selection strategies.
Authors: Jerry Huang, Prasanna Parthasarathi, Mehdi Rezagholizadeh, Sarath Chandar
Abstract: Large language models (LLMs) show an innate skill for solving language based tasks. But insights have suggested an inability to adjust for information or task-solving skills becoming outdated, as their knowledge, stored directly within their parameters, remains static in time. Tool use helps by offloading work to systems that the LLM can access through an interface, but LLMs that use them still must adapt to nonstationary environments for prolonged use, as new tools can emerge and existing tools can change. Nevertheless, tools require less specialized knowledge, therefore we hypothesize they are better suited for continual learning (CL) as they rely less on parametric memory for solving tasks and instead focus on learning when to apply pre-defined tools. To verify this, we develop a synthetic benchmark and follow this by aggregating existing NLP tasks to form a more realistic testing scenario. While we demonstrate scaling model size is not a solution, regardless of tool usage, continual learning techniques can enable tool LLMs to both adapt faster while forgetting less, highlighting their potential as continual learners.
Authors: Arav Agarwal, Karthik Mittal, Aidan Doyle, Pragnya Sridhar, Zipiao Wan, Jacob Arthur Doughty, Jaromir Savelka, Majd Sakr
Abstract: We conduct a preliminary study of the effect of GPT's temperature parameter on the diversity of GPT4-generated questions. We find that using higher temperature values leads to significantly higher diversity, with different temperatures exposing different types of similarity between generated sets of questions. We also demonstrate that diverse question generation is especially difficult for questions targeting lower levels of Bloom's Taxonomy.
Authors: Zoey Liu, Bonnie J. Dorr
Abstract: Recent work to enhance data partitioning strategies for more realistic model evaluation face challenges in providing a clear optimal choice. This study addresses these challenges, focusing on morphological segmentation and synthesizing limitations related to language diversity, adoption of multiple datasets and splits, and detailed model comparisons. Our study leverages data from 19 languages, including ten indigenous or endangered languages across 10 language families with diverse morphological systems (polysynthetic, fusional, and agglutinative) and different degrees of data availability. We conduct large-scale experimentation with varying sized combinations of training and evaluation sets as well as new test data. Our results show that, when faced with new test data: (1) models trained from random splits are able to achieve higher numerical scores; (2) model rankings derived from random splits tend to generalize more consistently.
Authors: Ryan Cotterell, Kevin Duh
Abstract: Low-resource named entity recognition is still an open problem in NLP. Most state-of-the-art systems require tens of thousands of annotated sentences in order to obtain high performance. However, for most of the world's languages, it is unfeasible to obtain such annotation. In this paper, we present a transfer learning scheme, whereby we train character-level neural CRFs to predict named entities for both high-resource languages and low resource languages jointly. Learning character representations for multiple related languages allows transfer among the languages, improving F1 by up to 9.8 points over a loglinear CRF baseline.
Authors: Xinwei Chen, Kun Li, Tianyou Song, Jiangjian Guo
Abstract: StackOverflow, with its vast question repository and limited labeled examples, raise an annotation challenge for us. We address this gap by proposing RoBERTa+MAML, a few-shot named entity recognition (NER) method leveraging meta-learning. Our approach, evaluated on the StackOverflow NER corpus (27 entity types), achieves a 5% F1 score improvement over the baseline. We improved the results further domain-specific phrase processing enhance results.
Authors: Jie Zhou, Xin Chen, Hang Zhang, Zhe Li
Abstract: In this paper, we explore the application of cognitive intelligence in legal knowledge, focusing on the development of judicial artificial intelligence. Utilizing natural language processing (NLP) as the core technology, we propose a method for the automatic construction of case knowledge graphs for judicial cases. Our approach centers on two fundamental NLP tasks: entity recognition and relationship extraction. We compare two pre-trained models for entity recognition to establish their efficacy. Additionally, we introduce a multi-task semantic relationship extraction model that incorporates translational embedding, leading to a nuanced contextualized case knowledge representation. Specifically, in a case study involving a "Motor Vehicle Traffic Accident Liability Dispute," our approach significantly outperforms the baseline model. The entity recognition F1 score improved by 0.36, while the relationship extraction F1 score increased by 2.37. Building on these results, we detail the automatic construction process of case knowledge graphs for judicial cases, enabling the assembly of knowledge graphs for hundreds of thousands of judgments. This framework provides robust semantic support for applications of judicial AI, including the precise categorization and recommendation of related cases.
Authors: Kyubyung Chae, Jaepill Choi, Yohan Jo, Taesup Kim
Abstract: A primary challenge in abstractive summarization is hallucination -- the phenomenon where a model generates plausible text that is absent in the source text. We hypothesize that the domain (or topic) of the source text triggers the model to generate text that is highly probable in the domain, neglecting the details of the source text. To alleviate this model bias, we introduce a decoding strategy based on domain-conditional pointwise mutual information. This strategy adjusts the generation probability of each token by comparing it with the token's marginal probability within the domain of the source text. According to evaluation on the XSUM dataset, our method demonstrates improvement in terms of faithfulness and source relevance. The code is publicly available at \url{https://github.com/qqplot/dcpmi}.
Authors: Kaixin Li, Yuchen Tian, Qisheng Hu, Ziyang Luo, Jing Ma
Abstract: Programming often involves converting detailed and complex specifications into code, a process during which developers typically utilize visual aids to more effectively convey concepts. While recent developments in Large Multimodal Models have demonstrated remarkable abilities in visual reasoning and mathematical tasks, there is little work on investigating whether these models can effectively interpret visual elements for code generation. To this end, we present MMCode, the first multi-modal coding dataset for evaluating algorithmic problem-solving skills in visually rich contexts. MMCode contains 3,548 questions and 6,620 images collected from real-world programming challenges harvested from 10 code competition websites, presenting significant challenges due to the extreme demand for reasoning abilities. Our experiment results show that current state-of-the-art models struggle to solve these problems. The results highlight the lack of powerful vision-code models, and we hope MMCode can serve as an inspiration for future works in this domain. The data and code are publicly available at https://github.com/happylkx/MMCode.
Authors: Yangyifan Xu, Jinliang Lu, Jiajun Zhang
Abstract: Ensembling different large language models (LLMs) to unleash their complementary potential and harness their individual strengths is highly valuable. Nevertheless, vocabulary discrepancies among various LLMs have constrained previous studies to either selecting or blending completely generated outputs. This limitation hinders the dynamic correction and enhancement of outputs during the generation process, resulting in a limited capacity for effective ensemble. To address this issue, we propose a novel method to Ensemble LLMs via Vocabulary Alignment (EVA). EVA bridges the lexical gap among various LLMs, enabling meticulous ensemble at each generation step. Specifically, we first learn mappings between the vocabularies of different LLMs with the assistance of overlapping tokens. Subsequently, these mappings are employed to project output distributions of LLMs into a unified space, facilitating a fine-grained ensemble. Finally, we design a filtering strategy to exclude models that generate unfaithful tokens. Experimental results on commonsense reasoning, arithmetic reasoning, machine translation, and data-to-text generation tasks demonstrate the superiority of our approach compared with individual LLMs and previous ensemble methods conducted on complete outputs. Further analyses confirm that our approach can leverage knowledge from different language models and yield consistent improvement.
Authors: Sergio Burdisso, Dairazalia S\'anchez-Cort\'es, Esa\'u Villatoro-Tello, Petr Motlicek
Abstract: Evaluating the reliability of news sources is a routine task for journalists and organizations committed to acquiring and disseminating accurate information. Recent research has shown that predicting sources' reliability represents an important first-prior step in addressing additional challenges such as fake news detection and fact-checking. In this paper, we introduce a novel approach for source reliability estimation that leverages reinforcement learning strategies for estimating the reliability degree of news sources. Contrary to previous research, our proposed approach models the problem as the estimation of a reliability degree, and not a reliability label, based on how all the news media sources interact with each other on the Web. We validated the effectiveness of our method on a news media reliability dataset that is an order of magnitude larger than comparable existing datasets. Results show that the estimated reliability degrees strongly correlates with journalists-provided scores (Spearman=0.80) and can effectively predict reliability labels (macro-avg. F$_1$ score=81.05). We release our implementation and dataset, aiming to provide a valuable resource for the NLP community working on information verification.
Authors: Jumbly Grindrod
Abstract: Do large language models like Chat-GPT or LLaMa meaningfully use the words they produce? Or are they merely clever prediction machines, simulating language use by producing statistically plausible text? There have already been some initial attempts to answer this question by showing that these models meet the criteria for entering meaningful states according to metasemantic theories of mental content. In this paper, I will argue for a different approach - that we should instead consider whether language models meet the criteria given by our best metasemantic theories of linguistic content. In that vein, I will illustrate how this can be done by applying two such theories to the case of language models: Gareth Evans' (1982) account of naming practices and Ruth Millikan's (1984, 2004, 2005) teleosemantics. In doing so, I will argue that it is a mistake to think that the failure of LLMs to meet plausible conditions for mental intentionality thereby renders their outputs meaningless, and that a distinguishing feature of linguistic intentionality - dependency on a pre-existing linguistic system - allows for the plausible result LLM outputs are meaningful.
Authors: Jumbly Grindrod
Abstract: The transformer architecture, introduced by Vaswani et al. (2017), is at the heart of the remarkable recent progress in the development of language models, including famous chatbots such as Chat-gpt and Bard. In this paper, I argue that we an extract from the way the transformer architecture works a picture of the relationship between context and meaning. I call this the transformer picture, and I argue that it is a novel with regard to two related philosophical debates: the contextualism debate regarding the extent of context-sensitivity across natural language, and the polysemy debate regarding how polysemy should be captured within an account of word meaning. Although much of the paper merely tries to position the transformer picture with respect to these two debates, I will also begin to make the case for the transformer picture.
Authors: Jumbly Grindrod
Abstract: This paper argues that large language models have a valuable scientific role to play in serving as scientific models of a language. Linguistic study should not only be concerned with the cognitive processes behind linguistic competence, but also with language understood as an external, social entity. Once this is recognized, the value of large language models as scientific models becomes clear. This paper defends this position against a number of arguments to the effect that language models provide no linguistic insight. It also draws upon recent work in philosophy of science to show how large language models could serve as scientific models.
Authors: Zepeng Ding, Wenhao Huang, Jiaqing Liang, Deqing Yang, Yanghua Xiao
Abstract: Relation triple extraction, which outputs a set of triples from long sentences, plays a vital role in knowledge acquisition. Large language models can accurately extract triples from simple sentences through few-shot learning or fine-tuning when given appropriate instructions. However, they often miss out when extracting from complex sentences. In this paper, we design an evaluation-filtering framework that integrates large language models with small models for relational triple extraction tasks. The framework includes an evaluation model that can extract related entity pairs with high precision. We propose a simple labeling principle and a deep neural network to build the model, embedding the outputs as prompts into the extraction process of the large model. We conduct extensive experiments to demonstrate that the proposed method can assist large language models in obtaining more accurate extraction results, especially from complex sentences containing multiple relational triples. Our evaluation model can also be embedded into traditional extraction models to enhance their extraction precision from complex sentences.
Authors: Matti Wiegmann, Jennifer Rakete, Magdalena Wolska, Benno Stein, Martin Potthast
Abstract: Trigger warnings are labels that preface documents with sensitive content if this content could be perceived as harmful by certain groups of readers. Since warnings about a document intuitively need to be shown before reading it, authors usually assign trigger warnings at the document level. What parts of their writing prompted them to assign a warning, however, remains unclear. We investigate for the first time the feasibility of identifying the triggering passages of a document, both manually and computationally. We create a dataset of 4,135 English passages, each annotated with one of eight common trigger warnings. In a large-scale evaluation, we then systematically evaluate the effectiveness of fine-tuned and few-shot classifiers, and their generalizability. We find that trigger annotation belongs to the group of subjective annotation tasks in NLP, and that automatic trigger classification remains challenging but feasible.
Authors: Juhwan Choi, Jungmin Yun, Kyohoon Jin, YoungBin Kim
Abstract: The quality of the dataset is crucial for ensuring optimal performance and reliability of downstream task models. However, datasets often contain noisy data inadvertently included during the construction process. Numerous attempts have been made to correct this issue through human annotators. However, hiring and managing human annotators is expensive and time-consuming. As an alternative, recent studies are exploring the use of large language models (LLMs) for data annotation. In this study, we present a case study that extends the application of LLM-based data annotation to enhance the quality of existing datasets through a cleansing strategy. Specifically, we leverage approaches such as chain-of-thought (CoT) and majority voting to imitate human annotation and classify unrelated documents from the Multi-News dataset, which is widely used for the multi-document summarization task. Through our proposed cleansing method, we introduce an enhanced Multi-News+. By employing LLMs for data cleansing, we demonstrate an efficient and effective approach to improving dataset quality without relying on expensive human annotation efforts.
Authors: Nailia Mirzakhmedova, Marcel Gohsen, Chia Hao Chang, Benno Stein
Abstract: Evaluating the quality of arguments is a crucial aspect of any system leveraging argument mining. However, it is a challenge to obtain reliable and consistent annotations regarding argument quality, as this usually requires domain-specific expertise of the annotators. Even among experts, the assessment of argument quality is often inconsistent due to the inherent subjectivity of this task. In this paper, we study the potential of using state-of-the-art large language models (LLMs) as proxies for argument quality annotators. To assess the capability of LLMs in this regard, we analyze the agreement between model, human expert, and human novice annotators based on an established taxonomy of argument quality dimensions. Our findings highlight that LLMs can produce consistent annotations, with a moderately high agreement with human experts across most of the quality dimensions. Moreover, we show that using LLMs as additional annotators can significantly improve the agreement between annotators. These results suggest that LLMs can serve as a valuable tool for automated argument quality assessment, thus streamlining and accelerating the evaluation of large argument datasets.
Authors: Hyunsoo Cho
Abstract: Many recent studies endeavor to improve open-source language models through imitation learning, and re-training on the synthetic instruction data from state-of-the-art proprietary models like ChatGPT and GPT-4. However, the innate nature of synthetic data inherently contains noisy data, giving rise to a substantial presence of low-quality data replete with erroneous responses, and flawed reasoning. Although we intuitively grasp the potential harm of noisy data, we lack a quantitative understanding of its impact. To this end, this paper explores the correlation between the degree of noise and its impact on language models through instruction tuning. We first introduce the Falsity-Controllable (FACO) dataset, which comprises pairs of true answers with corresponding reasoning, as well as false pairs to manually control the falsity ratio of the dataset.Through our extensive experiments, we found multiple intriguing findings of the correlation between the factuality of the dataset and instruction tuning: Specifically, we verified falsity of the instruction is highly relevant to various benchmark scores. Moreover, when LLMs are trained with false instructions, they learn to lie and generate fake unfaithful answers, even though they know the correct answer for the user request. Additionally, we noted that once the language model is trained with a dataset contaminated by noise, restoring its original performance is possible, but it failed to reach full performance.
Authors: Nicolas Wagner, Dongyang Fan, Martin Jaggi
Abstract: We explore on-device self-supervised collaborative fine-tuning of large language models with limited local data availability. Taking inspiration from the collaborative learning community, we introduce three distinct trust-weighted gradient aggregation schemes: weight similarity-based, prediction similarity-based and validation performance-based. To minimize communication overhead, we integrate Low-Rank Adaptation (LoRA) and only exchange LoRA weight updates. Our protocols, driven by prediction and performance metrics, surpass both FedAvg and local fine-tuning methods, which is particularly evident in realistic scenarios with more diverse local data distributions. The results underscore the effectiveness of our approach in addressing heterogeneity and scarcity within local datasets.
Authors: Bin Wang, Chengwei Wei, Zhengyuan Liu, Geyu Lin, Nancy F. Chen
Abstract: As the rapidly advancing domain of natural language processing (NLP), large language models (LLMs) have emerged as powerful tools for interpreting human commands and generating text across various tasks. Nonetheless, the resilience of LLMs to handle text containing inherent errors, stemming from human interactions and collaborative systems, has not been thoroughly explored. Our study investigates the resilience of LLMs against five common types of disruptions including 1) ASR (Automatic Speech Recognition) errors, 2) OCR (Optical Character Recognition) errors, 3) grammatical mistakes, 4) typographical errors, and 5) distractive content. We aim to investigate how these models react by deliberately embedding these errors into instructions. Our findings reveal that while some LLMs show a degree of resistance to certain types of noise, their overall performance significantly suffers. This emphasizes the importance of further investigation into enhancing model resilience. In response to the observed decline in performance, our study also evaluates a "re-pass" strategy, designed to purify the instructions of noise before the LLMs process them. Our analysis indicates that correcting noisy instructions, particularly for open-source LLMs, presents significant challenges.
Authors: Avinash Anand, Mohit Gupta, Kritarth Prasad, Ujjwal Goel, Naman Lal, Astha Verma, Rajiv Ratn Shah
Abstract: Citation Text Generation (CTG) is a task in natural language processing (NLP) that aims to produce text that accurately cites or references a cited document within a source document. In CTG, the generated text draws upon contextual cues from both the source document and the cited paper, ensuring accurate and relevant citation information is provided. Previous work in the field of citation generation is mainly based on the text summarization of documents. Following this, this paper presents a framework, and a comparative study to demonstrate the use of Large Language Models (LLMs) for the task of citation generation. Also, we have shown the improvement in the results of citation generation by incorporating the knowledge graph relations of the papers in the prompt for the LLM to better learn the relationship between the papers. To assess how well our model is performing, we have used a subset of standard S2ORC dataset, which only consists of computer science academic research papers in the English Language. Vicuna performs best for this task with 14.15 Meteor, 12.88 Rouge-1, 1.52 Rouge-2, and 10.94 Rouge-L. Also, Alpaca performs best, and improves the performance by 36.98% in Rouge-1, and 33.14% in Meteor by including knowledge graphs.
Authors: David Nadeau, Mike Kroutikov, Karen McNeil, Simon Baribeau
Abstract: This paper introduces fourteen novel datasets for the evaluation of Large Language Models' safety in the context of enterprise tasks. A method was devised to evaluate a model's safety, as determined by its ability to follow instructions and output factual, unbiased, grounded, and appropriate content. In this research, we used OpenAI GPT as point of comparison since it excels at all levels of safety. On the open-source side, for smaller models, Meta Llama2 performs well at factuality and toxicity but has the highest propensity for hallucination. Mistral hallucinates the least but cannot handle toxicity well. It performs well in a dataset mixing several tasks and safety vectors in a narrow vertical domain. Gemma, the newly introduced open-source model based on Google Gemini, is generally balanced but trailing behind. When engaging in back-and-forth conversation (multi-turn prompts), we find that the safety of open-source models degrades significantly. Aside from OpenAI's GPT, Mistral is the only model that still performed well in multi-turn tests.
Authors: Yang Gao, Dana Alon, Donald Metzler
Abstract: A key requirement in developing Generative Language Models (GLMs) is to have their values aligned with human values. Preference-based alignment is a widely used paradigm for this purpose, in which preferences over generation pairs are first elicited from human annotators or AI systems, and then fed into some alignment techniques, e.g., Direct Preference Optimization. However, a substantial percent (20 - 40%) of the preference pairs used in GLM alignment are noisy, and it remains unclear how the noise affects the alignment performance and how to mitigate its negative impact. In this paper, we propose a framework to inject desirable amounts and types of noise to the preferences, and systematically study the impact of preference noise on the alignment performance in two tasks (summarization and dialogue generation). We find that the alignment performance can be highly sensitive to the noise rates in the preference data: e.g., a 10 percentage points (pp) increase of the noise rate can lead to 30 pp drop in the alignment performance (in win rate). To mitigate the impact of noise, confidence-based data filtering shows significant benefit when certain types of noise are present. We hope our work can help the community better understand and mitigate the impact of preference noise in GLM alignment.
Authors: Yuchen Shi, Deqing Yang, Jingping Liu, Yanghua Xiao, Zongyu Wang, Huimin Xu
Abstract: Previous works of negation understanding mainly focus on negation cue detection and scope resolution, without identifying negation subject which is also significant to the downstream tasks. In this paper, we propose a new negation triplet extraction (NTE) task which aims to extract negation subject along with negation cue and scope. To achieve NTE, we devise a novel Syntax&Semantic-Enhanced Negation Extraction model, namely SSENE, which is built based on a generative pretrained language model (PLM) {of Encoder-Decoder architecture} with a multi-task learning framework. Specifically, the given sentence's syntactic dependency tree is incorporated into the PLM's encoder to discover the correlations between the negation subject, cue and scope. Moreover, the semantic consistency between the sentence and the extracted triplet is ensured by an auxiliary task learning. Furthermore, we have constructed a high-quality Chinese dataset NegComment based on the users' reviews from the real-world platform of Meituan, upon which our evaluations show that SSENE achieves the best NTE performance compared to the baselines. Our ablation and case studies also demonstrate that incorporating the syntactic information helps the PLM's recognize the distant dependency between the subject and cue, and the auxiliary task learning is helpful to extract the negation triplets with more semantic consistency.
Authors: Yuxi Li, Yi Liu, Gelei Deng, Ying Zhang, Wenjia Song, Ling Shi, Kailong Wang, Yuekang Li, Yang Liu, Haoyu Wang
Abstract: With the expanding application of Large Language Models (LLMs) in various domains, it becomes imperative to comprehensively investigate their unforeseen behaviors and consequent outcomes. In this study, we introduce and systematically explore the phenomenon of "glitch tokens", which are anomalous tokens produced by established tokenizers and could potentially compromise the models' quality of response. Specifically, we experiment on seven top popular LLMs utilizing three distinct tokenizers and involving a totally of 182,517 tokens. We present categorizations of the identified glitch tokens and symptoms exhibited by LLMs when interacting with glitch tokens. Based on our observation that glitch tokens tend to cluster in the embedding space, we propose GlitchHunter, a novel iterative clustering-based technique, for efficient glitch token detection. The evaluation shows that our approach notably outperforms three baseline methods on eight open-source LLMs. To the best of our knowledge, we present the first comprehensive study on glitch tokens. Our new detection further provides valuable insights into mitigating tokenization-related errors in LLMs.
Authors: Sanxing Chen, Sam Wiseman, Bhuwan Dhingra
Abstract: The desire and ability to seek new information strategically are fundamental to human learning but often overlooked in current language agent development. Using a web shopping task as an example, we show that it can be reformulated and solved as a retrieval task without a requirement of interactive information seeking. We then redesign the task to introduce a new role of shopper, serving as a realistically constrained communication channel. The agents in our proposed ChatShop task explore user preferences in open-ended conversation to make informed decisions. Our experiments demonstrate that the proposed task can effectively evaluate the agent's ability to explore and gradually accumulate information through multi-turn interaction. We also show that LLM-simulated shoppers serve as a good proxy to real human shoppers and discover similar error patterns of agents.
Authors: Yuzhen Huang, Jinghan Zhang, Zifei Shan, Junxian He
Abstract: There is a belief that learning to compress well will lead to intelligence. Recently, language modeling has been shown to be equivalent to compression, which offers a compelling rationale for the success of large language models (LLMs): the development of more advanced language models is essentially enhancing compression which facilitates intelligence. Despite such appealing discussions, little empirical evidence is present for the interplay between compression and intelligence. In this work, we examine their relationship in the context of LLMs, treating LLMs as data compressors. Given the abstract concept of "intelligence", we adopt the average downstream benchmark scores as a surrogate, specifically targeting intelligence related to knowledge and commonsense, coding, and mathematical reasoning. Across 12 benchmarks, our study brings together 30 public LLMs that originate from diverse organizations. Remarkably, we find that LLMs' intelligence -- reflected by average benchmark scores -- almost linearly correlates with their ability to compress external text corpora. These results provide concrete evidence supporting the belief that superior compression indicates greater intelligence. Furthermore, our findings suggest that compression efficiency, as an unsupervised metric derived from raw text corpora, serves as a reliable evaluation measure that is linearly associated with the model capabilities. We open-source our compression datasets as well as our data collection pipelines to facilitate future researchers to assess compression properly.
Authors: Adi Simhi, Jonathan Herzig, Idan Szpektor, Yonatan Belinkov
Abstract: Large language models (LLMs) are susceptible to hallucination, which sparked a widespread effort to detect and prevent them. Recent work attempts to mitigate hallucinations by intervening in the model's computation during generation, using different setups and heuristics. Those works lack separation between different hallucination causes. In this work, we first introduce an approach for constructing datasets based on the model knowledge for detection and intervention methods in closed-book and open-book question-answering settings. We then characterize the effect of different choices for intervention, such as the intervened components (MLPs, attention block, residual stream, and specific heads), and how often and how strongly to intervene. We find that intervention success varies depending on the component, with some components being detrimental to language modeling capabilities. Finally, we find that interventions can benefit from pre-hallucination steering direction instead of post-hallucination. The code is available at https://github.com/technion-cs-nlp/hallucination-mitigation
URLs: https://github.com/technion-cs-nlp/hallucination-mitigation
Authors: Clemencia Siro, Mohammad Aliannejadi, Maarten de Rijke
Abstract: Crowdsourced labels play a crucial role in evaluating task-oriented dialogue systems (TDSs). Obtaining high-quality and consistent ground-truth labels from annotators presents challenges. When evaluating a TDS, annotators must fully comprehend the dialogue before providing judgments. Previous studies suggest using only a portion of the dialogue context in the annotation process. However, the impact of this limitation on label quality remains unexplored. This study investigates the influence of dialogue context on annotation quality, considering the truncated context for relevance and usefulness labeling. We further propose to use large language models (LLMs) to summarize the dialogue context to provide a rich and short description of the dialogue context and study the impact of doing so on the annotator's performance. Reducing context leads to more positive ratings. Conversely, providing the entire dialogue context yields higher-quality relevance ratings but introduces ambiguity in usefulness ratings. Using the first user utterance as context leads to consistent ratings, akin to those obtained using the entire dialogue, with significantly reduced annotation effort. Our findings show how task design, particularly the availability of dialogue context, affects the quality and consistency of crowdsourced evaluation labels.
Authors: Hang Gao, Yongfeng Zhang
Abstract: In the realm of artificial intelligence, the adaptation of Large Language Model (LLM)-based agents to execute tasks via natural language prompts represents a significant advancement, notably eliminating the need for explicit retraining or fine tuning for fixed-answer tasks such as common sense questions and yes/no queries. However, the application of In-context Learning to open-ended challenges, such as poetry creation, reveals substantial limitations due to the comprehensiveness of the provided examples and agent's ability to understand the content expressed in the problem, leading to outputs that often diverge significantly from expected results. Addressing this gap, our study introduces the Memory-Sharing (MS) framework for LLM multi-agents, which utilizes a real-time memory storage and retrieval system to enhance the In-context Learning process. Each "memory" within this system captures both the posed query and the corresponding real-time response from an LLM-based agent, aggregating these memories from a broad spectrum of similar agents to enrich the memory pool shared by all agents. This framework not only aids agents in identifying the most relevant examples for specific tasks but also evaluates the potential utility of their memories for future applications by other agents. Empirical validation across three distinct domains involving specialized functions of agents demonstrates that the MS framework significantly improve the agent's performance regrading the open-ended questions. Furthermore, we also discuss what type of memory pool and what retrieval strategy in MS can better help agents, offering a future develop direction of MS. The code and data are available at: https://github.com/GHupppp/MemorySharingLLM
Authors: Roberto Bigazzi, Federico Landi, Marcella Cornia, Silvia Cascianelli, Lorenzo Baraldi, Rita Cucchiara
Abstract: Embodied AI has been recently gaining attention as it aims to foster the development of autonomous and intelligent agents. In this paper, we devise a novel embodied setting in which an agent needs to explore a previously unknown environment while recounting what it sees during the path. In this context, the agent needs to navigate the environment driven by an exploration goal, select proper moments for description, and output natural language descriptions of relevant objects and scenes. Our model integrates a novel self-supervised exploration module with penalty, and a fully-attentive captioning model for explanation. Also, we investigate different policies for selecting proper moments for explanation, driven by information coming from both the environment and the navigation. Experiments are conducted on photorealistic environments from the Matterport3D dataset and investigate the navigation and explanation capabilities of the agent as well as the role of their interactions.
Authors: Roberto Bigazzi, Marcella Cornia, Silvia Cascianelli, Lorenzo Baraldi, Rita Cucchiara
Abstract: The development of embodied agents that can communicate with humans in natural language has gained increasing interest over the last years, as it facilitates the diffusion of robotic platforms in human-populated environments. As a step towards this objective, in this work, we tackle a setting for visual navigation in which an autonomous agent needs to explore and map an unseen indoor environment while portraying interesting scenes with natural language descriptions. To this end, we propose and evaluate an approach that combines recent advances in visual robotic exploration and image captioning on images generated through agent-environment interaction. Our approach can generate smart scene descriptions that maximize semantic knowledge of the environment and avoid repetitions. Further, such descriptions offer user-understandable insights into the robot's representation of the environment by highlighting the prominent objects and the correlation between them as encountered during the exploration. To quantitatively assess the performance of the proposed approach, we also devise a specific score that takes into account both exploration and description skills. The experiments carried out on both photorealistic simulated environments and real-world ones demonstrate that our approach can effectively describe the robot's point of view during exploration, improving the human-friendly interpretability of its observations.
Authors: Silvia Garc\'ia-M\'endez, Milagros Fern\'andez-Gavilanes, Jonathan Juncal-Mart\'inez, Francisco J. Gonz\'alez-Casta\~no, Oscar Barba Seara
Abstract: Short texts are omnipresent in real-time news, social network commentaries, etc. Traditional text representation methods have been successfully applied to self-contained documents of medium size. However, information in short texts is often insufficient, due, for example, to the use of mnemonics, which makes them hard to classify. Therefore, the particularities of specific domains must be exploited. In this article we describe a novel system that combines Natural Language Processing techniques with Machine Learning algorithms to classify banking transaction descriptions for personal finance management, a problem that was not previously considered in the literature. We trained and tested that system on a labelled dataset with real customer transactions that will be available to other researchers on request. Motivated by existing solutions in spam detection, we also propose a short text similarity detector to reduce training set size based on the Jaccard distance. Experimental results with a two-stage classifier combining this detector with a SVM indicate a high accuracy in comparison with alternative approaches, taking into account complexity and computing time. Finally, we present a use case with a personal finance application, CoinScrap, which is available at Google Play and App Store.
Authors: Silvia Garc\'ia-M\'endez, Francisco de Arriba-P\'erez, Ana Barros-Vila, Francisco J. Gonz\'alez-Casta\~no
Abstract: Microblogging platforms, of which Twitter is a representative example, are valuable information sources for market screening and financial models. In them, users voluntarily provide relevant information, including educated knowledge on investments, reacting to the state of the stock markets in real-time and, often, influencing this state. We are interested in the user forecasts in financial, social media messages expressing opportunities and precautions about assets. We propose a novel Targeted Aspect-Based Emotion Analysis (TABEA) system that can individually discern the financial emotions (positive and negative forecasts) on the different stock market assets in the same tweet (instead of making an overall guess about that whole tweet). It is based on Natural Language Processing (NLP) techniques and Machine Learning streaming algorithms. The system comprises a constituency parsing module for parsing the tweets and splitting them into simpler declarative clauses; an offline data processing module to engineer textual, numerical and categorical features and analyse and select them based on their relevance; and a stream classification module to continuously process tweets on-the-fly. Experimental results on a labelled data set endorse our solution. It achieves over 90% precision for the target emotions, financial opportunity, and precaution on Twitter. To the best of our knowledge, no prior work in the literature has addressed this problem despite its practical interest in decision-making, and we are not aware of any previous NLP nor online Machine Learning approaches to TABEA.
Authors: Hwiyeol Jo, Taiwoo Park, Nayoung Choi, Changbong Kim, Ohjoon Kwon, Donghyeon Jeon, Hyunwoo Lee, Eui-Hyeon Lee, Kyoungho Shin, Sun Suk Lim, Kyungmi Kim, Jihye Lee, Sun Kim
Abstract: Although there has been a growing interest among industries to integrate generative LLMs into their services, limited experiences and scarcity of resources acts as a barrier in launching and servicing large-scale LLM-based conversational services. In this paper, we share our experiences in developing and operating generative AI models within a national-scale search engine, with a specific focus on the sensitiveness of user queries. We propose a taxonomy for sensitive search queries, outline our approaches, and present a comprehensive analysis report on sensitive queries from actual users.
Authors: Yabin Zhang, Wenhui Yu, Erhan Zhang, Xu Chen, Lantao Hu, Peng Jiang, Kun Gai
Abstract: ChatGPT has achieved remarkable success in natural language understanding. Considering that recommendation is indeed a conversation between users and the system with items as words, which has similar underlying pattern with ChatGPT, we design a new chat framework in item index level for the recommendation task. Our novelty mainly contains three parts: model, training and inference. For the model part, we adopt Generative Pre-training Transformer (GPT) as the sequential recommendation model and design a user modular to capture personalized information. For the training part, we adopt the two-stage paradigm of ChatGPT, including pre-training and fine-tuning. In the pre-training stage, we train GPT model by auto-regression. In the fine-tuning stage, we train the model with prompts, which include both the newly-generated results from the model and the user's feedback. For the inference part, we predict several user interests as user representations in an autoregressive manner. For each interest vector, we recall several items with the highest similarity and merge the items recalled by all interest vectors into the final result. We conduct experiments with both offline public datasets and online A/B test to demonstrate the effectiveness of our proposed method.
Authors: Xiaoteng Shen, Rui Zhang, Xiaoyan Zhao, Jieming Zhu, Xi Xiao
Abstract: The emergence of large language models (LLMs) has revolutionized the capabilities of text comprehension and generation. Multi-modal generation attracts great attention from both the industry and academia, but there is little work on personalized generation, which has important applications such as recommender systems. This paper proposes the first method for personalized multimodal generation using LLMs, showcases its applications and validates its performance via an extensive experimental study on two datasets. The proposed method, Personalized Multimodal Generation (PMG for short) first converts user behaviors (e.g., clicks in recommender systems or conversations with a virtual assistant) into natural language to facilitate LLM understanding and extract user preference descriptions. Such user preferences are then fed into a generator, such as a multimodal LLM or diffusion model, to produce personalized content. To capture user preferences comprehensively and accurately, we propose to let the LLM output a combination of explicit keywords and implicit embeddings to represent user preferences. Then the combination of keywords and embeddings are used as prompts to condition the generator. We optimize a weighted sum of the accuracy and preference scores so that the generated content has a good balance between them. Compared to a baseline method without personalization, PMG has a significant improvement on personalization for up to 8% in terms of LPIPS while retaining the accuracy of generation.
Authors: Shangyu Chen, Zibo Zhao, Yuanyuan Zhao, Xiang Li
Abstract: The emergence of Large Language Models (LLMs) has innovated the development of dialog agents. Specially, a well-trained LLM, as a central process unit, is capable of providing fluent and reasonable response for user's request. Besides, auxiliary tools such as external knowledge retrieval, personalized character for vivid response, short/long-term memory for ultra long context management are developed, completing the usage experience for LLM-based dialog agents. However, the above-mentioned techniques does not solve the issue of \textbf{personalization from user perspective}: agents response in a same fashion to different users, without consideration of their features, such as habits, interests and past experience. In another words, current implementation of dialog agents fail in ``knowing the user''. The capacity of well-description and representation of user is under development. In this work, we proposed a framework for dialog agent to incorporate user profiling (initialization, update): user's query and response is analyzed and organized into a structural user profile, which is latter served to provide personal and more precise response. Besides, we proposed a series of evaluation protocols for personalization: to what extend the response is personal to the different users. The framework is named as \method{}, inspired by inscription of ``Know Yourself'' in the temple of Apollo (also known as \method{}) in Ancient Greek. Few works have been conducted on incorporating personalization into LLM, \method{} is a pioneer work on guiding LLM's response to meet individuation via the application of dialog agents, with a set of evaluation methods for measurement in personalization.
Authors: Haokun Zhao, Haixia Han, Jie Shi, Chengyu Du, Jiaqing Liang, Yanghua Xiao
Abstract: Large Language Models (LLMs) demonstrate impressive performance in various downstream tasks. However, they may still generate incorrect responses in certain scenarios due to the knowledge deficiencies and the flawed pre-training data. Continual Learning (CL) is a commonly used method to address this issue. Traditional CL is task-oriented, using novel or factually accurate data to retrain LLMs from scratch. However, this method requires more task-related training data and incurs expensive training costs. To address this challenge, we propose the Continue Evolving from Mistakes (CEM) method, inspired by the 'summarize mistakes' learning skill, to achieve iterative refinement of LLMs. Specifically, the incorrect responses of LLMs indicate knowledge deficiencies related to the questions. Therefore, we collect corpora with these knowledge from multiple data sources and follow it up with iterative supplementary training for continuous, targeted knowledge updating and supplementation. Meanwhile, we developed two strategies to construct supplementary training sets to enhance the LLM's understanding of the corpus and prevent catastrophic forgetting. We conducted extensive experiments to validate the effectiveness of this CL method. In the best case, our method resulted in a 17.00\% improvement in the accuracy of the LLM.
Authors: Alexandre Audibert, Aur\'elien Gauffre, Massih-Reza Amini
Abstract: Learning an effective representation in multi-label text classification (MLTC) is a significant challenge in NLP. This challenge arises from the inherent complexity of the task, which is shaped by two key factors: the intricate connections between labels and the widespread long-tailed distribution of the data. To overcome this issue, one potential approach involves integrating supervised contrastive learning with classical supervised loss functions. Although contrastive learning has shown remarkable performance in multi-class classification, its impact in the multi-label framework has not been thoroughly investigated. In this paper, we conduct an in-depth study of supervised contrastive learning and its influence on representation in MLTC context. We emphasize the importance of considering long-tailed data distributions to build a robust representation space, which effectively addresses two critical challenges associated with contrastive learning that we identify: the "lack of positives" and the "attraction-repulsion imbalance". Building on this insight, we introduce a novel contrastive loss function for MLTC. It attains Micro-F1 scores that either match or surpass those obtained with other frequently employed loss functions, and demonstrates a significant improvement in Macro-F1 scores across three multi-label datasets.
Authors: Xiang Zhang, Khatoon Khedri, Reza Rawassizadeh
Abstract: Large Language Models (LLMs) can automate or substitute different types of tasks in the software engineering process. This study evaluates the resource utilization and accuracy of LLM in interpreting and executing natural language queries against traditional SQL within relational database management systems. We empirically examine the resource utilization and accuracy of nine LLMs varying from 7 to 34 Billion parameters, including Llama2 7B, Llama2 13B, Mistral, Mixtral, Optimus-7B, SUS-chat-34B, platypus-yi-34b, NeuralHermes-2.5-Mistral-7B and Starling-LM-7B-alpha, using a small transaction dataset. Our findings indicate that using LLMs for database queries incurs significant energy overhead (even small and quantized models), making it an environmentally unfriendly approach. Therefore, we advise against replacing relational databases with LLMs due to their substantial resource utilization.
Authors: Je-Yong Lee, Donghyun Lee, Genghan Zhang, Mo Tiwari, Azalia Mirhoseini
Abstract: Large Language Models (LLMs) have dramatically advanced AI applications, yet their deployment remains challenging due to their immense inference costs. Recent studies ameliorate the computational costs of LLMs by increasing their activation sparsity but suffer from significant performance degradation on downstream tasks. In this work, we introduce a new framework for sparsifying the activations of base LLMs and reducing inference costs, dubbed Contextually Aware Thresholding for Sparsity (CATS). CATS is relatively simple, easy to implement, and highly effective. At the heart of our framework is a new non-linear activation function. We demonstrate that CATS can be applied to various base models, including Mistral-7B and Llama2-7B, and outperforms existing sparsification techniques in downstream task performance. More precisely, CATS-based models often achieve downstream task performance within 1-2% of their base models without any fine-tuning and even at activation sparsity levels of 50%. Furthermore, CATS-based models converge faster and display better task performance than competing techniques when fine-tuning is applied. Finally, we develop a custom GPU kernel for efficient implementation of CATS that translates the activation of sparsity of CATS to real wall-clock time speedups. Our custom kernel implementation of CATS results in a ~15% improvement in wall-clock inference latency of token generation on both Llama-7B and Mistral-7B.
Authors: Yingchaojie Feng, Zhizhang Chen, Zhining Kang, Sijia Wang, Minfeng Zhu, Wei Zhang, Wei Chen
Abstract: The proliferation of large language models (LLMs) has underscored concerns regarding their security vulnerabilities, notably against jailbreak attacks, where adversaries design jailbreak prompts to circumvent safety mechanisms for potential misuse. Addressing these concerns necessitates a comprehensive analysis of jailbreak prompts to evaluate LLMs' defensive capabilities and identify potential weaknesses. However, the complexity of evaluating jailbreak performance and understanding prompt characteristics makes this analysis laborious. We collaborate with domain experts to characterize problems and propose an LLM-assisted framework to streamline the analysis process. It provides automatic jailbreak assessment to facilitate performance evaluation and support analysis of components and keywords in prompts. Based on the framework, we design JailbreakLens, a visual analysis system that enables users to explore the jailbreak performance against the target model, conduct multi-level analysis of prompt characteristics, and refine prompt instances to verify findings. Through a case study, technical evaluations, and expert interviews, we demonstrate our system's effectiveness in helping users evaluate model security and identify model weaknesses.
Authors: Xuezhe Ma, Xiaomeng Yang, Wenhan Xiong, Beidi Chen, Lili Yu, Hao Zhang, Jonathan May, Luke Zettlemoyer, Omer Levy, Chunting Zhou
Abstract: The quadratic complexity and weak length extrapolation of Transformers limits their ability to scale to long sequences, and while sub-quadratic solutions like linear attention and state space models exist, they empirically underperform Transformers in pretraining efficiency and downstream task accuracy. We introduce Megalodon, a neural architecture for efficient sequence modeling with unlimited context length. Megalodon inherits the architecture of Mega (exponential moving average with gated attention), and further introduces multiple technical components to improve its capability and stability, including complex exponential moving average (CEMA), timestep normalization layer, normalized attention mechanism and pre-norm with two-hop residual configuration. In a controlled head-to-head comparison with Llama2, Megalodon achieves better efficiency than Transformer in the scale of 7 billion parameters and 2 trillion training tokens. Megalodon reaches a training loss of 1.70, landing mid-way between Llama2-7B (1.75) and 13B (1.67). Code: https://github.com/XuezheMax/megalodon
Authors: William Merrill, Jackson Petty, Ashish Sabharwal
Abstract: State-space models (SSMs) have emerged as a potential alternative architecture for building large language models (LLMs) compared to the previously ubiquitous transformer architecture. One theoretical weakness of transformers is that they cannot express certain kinds of sequential computation and state tracking (Merrill and Sabharwal, 2023), which SSMs are explicitly designed to address via their close architectural similarity to recurrent neural networks (RNNs). But do SSMs truly have an advantage (over transformers) in expressive power for state tracking? Surprisingly, the answer is no. Our analysis reveals that the expressive power of SSMs is limited very similarly to transformers: SSMs cannot express computation outside the complexity class $\mathsf{TC}^0$. In particular, this means they cannot solve simple state-tracking problems like permutation composition. It follows that SSMs are provably unable to accurately track chess moves with certain notation, evaluate code, or track entities in a long narrative. To supplement our formal analysis, we report experiments showing that Mamba-style SSMs indeed struggle with state tracking. Thus, despite its recurrent formulation, the "state" in an SSM is an illusion: SSMs have similar expressiveness limitations to non-recurrent models like transformers, which may fundamentally limit their ability to solve real-world state-tracking problems.
Authors: Subhojyoti Mukherjee, Ge Liu, Aniket Deshmukh, Anusha Lalitha, Yifei Ma, Branislav Kveton
Abstract: Transduction, the ability to include query-specific examples in the prompt at inference time, is one of the emergent abilities of large language models (LLMs). In this work, we propose a framework for adaptive prompt design called active transductive inference (ATI). We design the LLM prompt by adaptively choosing few-shot examples for a given inference query. The examples are initially unlabeled and we query the user to label the most informative ones, which maximally reduces the uncertainty in the LLM prediction. We propose two algorithms, GO and SAL, which differ in how the few-shot examples are chosen. We analyze these algorithms in linear models: first GO and then use its equivalence with SAL. We experiment with many different tasks and show that GO and SAL outperform other methods for choosing few-shot examples in the LLM prompt at inference time.
Authors: Junjielong Xu, Ying Fu, Shin Hwei Tan, Pinjia He
Abstract: Large language models (LLMs) have achieved decent results on automated program repair (APR). However, the next token prediction training objective of decoder-only LLMs (e.g., GPT-4) is misaligned with the masked span prediction objective of current infilling-style methods, which impedes LLMs from fully leveraging pre-trained knowledge for program repair. In addition, while some LLMs are capable of locating and repairing bugs end-to-end when using the related artifacts (e.g., test cases) as input, existing methods regard them as separate tasks and ask LLMs to generate patches at fixed locations. This restriction hinders LLMs from exploring potential patches beyond the given locations. In this paper, we investigate a new approach to adapt LLMs to program repair. Our core insight is that LLM's APR capability can be greatly improved by simply aligning the output to their training objective and allowing them to refine the whole program without first performing fault localization. Based on this insight, we designed D4C, a straightforward prompting framework for APR. D4C can repair 180 bugs correctly in Defects4J, with each patch being sampled only 10 times. This surpasses the SOTA APR methods with perfect fault localization by 10% and reduces the patch sampling number by 90%. Our findings reveal that (1) objective alignment is crucial for fully exploiting LLM's pre-trained capability, and (2) replacing the traditional localize-then-repair workflow with direct debugging is more effective for LLM-based APR methods. Thus, we believe this paper introduces a new mindset for harnessing LLMs in APR.
Authors: Mengnan Qi, Yufan Huang, Yongqiang Yao, Maoquan Wang, Bin Gu, Neel Sundaresan
Abstract: Large language models (LLMs) has experienced exponential growth, they demonstrate remarkable performance across various tasks. Notwithstanding, contemporary research primarily centers on enhancing the size and quality of pretraining data, still utilizing the next token prediction task on autoregressive transformer model structure. The efficacy of this task in truly facilitating the model's comprehension of code logic remains questionable, we speculate that it still interprets code as mere text, while human emphasizes the underlying logical knowledge. In order to prove it, we introduce a new task, "Logically Equivalent Code Selection," which necessitates the selection of logically equivalent code from a candidate set, given a query code. Our experimental findings indicate that current LLMs underperform in this task, since they understand code by unordered bag of keywords. To ameliorate their performance, we propose an advanced pretraining task, "Next Token Prediction+". This task aims to modify the sentence embedding distribution of the LLM without sacrificing its generative capabilities. Our experimental results reveal that following this pretraining, both Code Llama and StarCoder, the prevalent code domain pretraining models, display significant improvements on our logically equivalent code selection task and the code completion task.
Authors: Henry Peng Zou, Gavin Heqing Yu, Ziwei Fan, Dan Bu, Han Liu, Peng Dai, Dongmei Jia, Cornelia Caragea
Abstract: In e-commerce, accurately extracting product attribute values from multimodal data is crucial for improving user experience and operational efficiency of retailers. However, previous approaches to multimodal attribute value extraction often struggle with implicit attribute values embedded in images or text, rely heavily on extensive labeled data, and can easily confuse similar attribute values. To address these issues, we introduce EIVEN, a data- and parameter-efficient generative framework that pioneers the use of multimodal LLM for implicit attribute value extraction. EIVEN leverages the rich inherent knowledge of a pre-trained LLM and vision encoder to reduce reliance on labeled data. We also introduce a novel Learning-by-Comparison technique to reduce model confusion by enforcing attribute value comparison and difference identification. Additionally, we construct initial open-source datasets for multimodal implicit attribute value extraction. Our extensive experiments reveal that EIVEN significantly outperforms existing methods in extracting implicit attribute values while requiring less labeled data.
Authors: Ayush Thakur, Raghav Gupta
Abstract: The relentless pursuit of enhancing Large Language Models (LLMs) has led to the advent of Super Retrieval-Augmented Generation (Super RAGs), a novel approach designed to elevate the performance of LLMs by integrating external knowledge sources with minimal structural modifications. This paper presents the integration of Super RAGs into the Mistral 8x7B v1, a state-of-the-art LLM, and examines the resultant improvements in accuracy, speed, and user satisfaction. Our methodology uses a fine-tuned instruct model setup and a cache tuning fork system, ensuring efficient and relevant data retrieval. The evaluation, conducted over several epochs, demonstrates significant enhancements across all metrics. The findings suggest that Super RAGs can effectively augment LLMs, paving the way for more sophisticated and reliable AI systems. This research contributes to the field by providing empirical evidence of the benefits of Super RAGs and offering insights into their potential applications.
Authors: Yuwei Tang, Zhenyi Lin, Qilong Wang, Pengfei Zhu, Qinghua Hu
Abstract: Recently, pre-trained vision-language models (e.g., CLIP) have shown great potential in few-shot learning and attracted a lot of research interest. Although efforts have been made to improve few-shot ability of CLIP, key factors on the effectiveness of existing methods have not been well studied, limiting further exploration of CLIP's potential in few-shot learning. In this paper, we first introduce a unified formulation to analyze CLIP-based few-shot learning methods from a perspective of logit bias, which encourages us to learn an effective logit bias for further improving performance of CLIP-based few-shot learning methods. To this end, we disassemble three key components involved in computation of logit bias (i.e., logit features, logit predictor, and logit fusion) and empirically analyze the effect on performance of few-shot classification. Based on analysis of key components, this paper proposes a novel AMU-Tuning method to learn effective logit bias for CLIP-based few-shot classification. Specifically, our AMU-Tuning predicts logit bias by exploiting the appropriate $\underline{\textbf{A}}$uxiliary features, which are fed into an efficient feature-initialized linear classifier with $\underline{\textbf{M}}$ulti-branch training. Finally, an $\underline{\textbf{U}}$ncertainty-based fusion is developed to incorporate logit bias into CLIP for few-shot classification. The experiments are conducted on several widely used benchmarks, and the results show AMU-Tuning clearly outperforms its counterparts while achieving state-of-the-art performance of CLIP-based few-shot learning without bells and whistles.
Authors: Benjue Weng
Abstract: With the surge of ChatGPT,the use of large models has significantly increased,rapidly rising to prominence across the industry and sweeping across the internet. This article is a comprehensive review of fine-tuning methods for large models. This paper investigates the latest technological advancements and the application of advanced methods in aspects such as task-adaptive fine-tuning,domain-adaptive fine-tuning,few-shot learning,knowledge distillation,multi-task learning,parameter-efficient fine-tuning,and dynamic fine-tuning.
Authors: Amit Finkman, Eden Bar-Kochva, Avishag Shapira, Dudu Mimran, Yuval Elovici, Asaf Shabtai
Abstract: LLM-based code assistants are becoming increasingly popular among developers. These tools help developers improve their coding efficiency and reduce errors by providing real-time suggestions based on the developer's codebase. While beneficial, these tools might inadvertently expose the developer's proprietary code to the code assistant service provider during the development process. In this work, we propose two complementary methods to mitigate the risk of code leakage when using LLM-based code assistants. The first is a technique for reconstructing a developer's original codebase from code segments sent to the code assistant service (i.e., prompts) during the development process, enabling assessment and evaluation of the extent of code leakage to third parties (or adversaries). The second is CodeCloak, a novel deep reinforcement learning agent that manipulates the prompts before sending them to the code assistant service. CodeCloak aims to achieve the following two contradictory goals: (i) minimizing code leakage, while (ii) preserving relevant and useful suggestions for the developer. Our evaluation, employing GitHub Copilot, StarCoder, and CodeLlama LLM-based code assistants models, demonstrates the effectiveness of our CodeCloak approach on a diverse set of code repositories of varying sizes, as well as its transferability across different models. In addition, we generate a realistic simulated coding environment to thoroughly analyze code leakage risks and evaluate the effectiveness of our proposed mitigation techniques under practical development scenarios.
Authors: Sanat Sharma, Jayant Kumar, Twisha Naik, Zhaoyu Lu, Arvind Srikantan, Tracy Holloway King
Abstract: Accurate explicit and implicit product identification in search queries is critical for enhancing user experiences, especially at a company like Adobe which has over 50 products and covers queries across hundreds of tools. In this work, we present a novel approach to training a product classifier from user behavioral data. Our semantic model led to >25% relative improvement in CTR (click through rate) across the deployed surfaces; a >50% decrease in null rate; a 2x increase in the app cards surfaced, which helps drive product visibility.
Authors: Dipendra Misra, Aldo Pacchiano, Robert E. Schapire
Abstract: We study interactive learning in a setting where the agent has to generate a response (e.g., an action or trajectory) given a context and an instruction. In contrast, to typical approaches that train the system using reward or expert supervision on response, we study learning with hindsight instruction where a teacher provides an instruction that is most suitable for the agent's generated response. This hindsight labeling of instruction is often easier to provide than providing expert supervision of the optimal response which may require expert knowledge or can be impractical to elicit. We initiate the theoretical analysis of interactive learning with hindsight labeling. We first provide a lower bound showing that in general, the regret of any algorithm must scale with the size of the agent's response space. We then study a specialized setting where the underlying instruction-response distribution can be decomposed as a low-rank matrix. We introduce an algorithm called LORIL for this setting and show that its regret scales as $\sqrt{T}$ where $T$ is the number of rounds and depends on the intrinsic rank but does not depend on the size of the agent's response space. We provide experiments in two domains showing that LORIL outperforms baselines even when the low-rank assumption is violated.
Authors: Jiang Li, Xiangdong Su, Yeyun Gong, Guanglai Gao
Abstract: Recent studies have highlighted the effectiveness of tensor decomposition methods in the Temporal Knowledge Graphs Embedding (TKGE) task. However, we found that inherent heterogeneity among factor tensors in tensor decomposition significantly hinders the tensor fusion process and further limits the performance of link prediction. To overcome this limitation, we introduce a novel method that maps factor tensors onto a unified smooth Lie group manifold to make the distribution of factor tensors approximating homogeneous in tensor decomposition. We provide the theoretical proof of our motivation that homogeneous tensors are more effective than heterogeneous tensors in tensor fusion and approximating the target for tensor decomposition based TKGE methods. The proposed method can be directly integrated into existing tensor decomposition based TKGE methods without introducing extra parameters. Extensive experiments demonstrate the effectiveness of our method in mitigating the heterogeneity and in enhancing the tensor decomposition based TKGE models.
Authors: Dongseong Hwang, Weiran Wang, Zhuoyuan Huo, Khe Chai Sim, Pedro Moreno Mengibar
Abstract: While Transformers have revolutionized deep learning, their quadratic attention complexity hinders their ability to process infinitely long inputs. We propose Feedback Attention Memory (FAM), a novel Transformer architecture that leverages a feedback loop to enable the network to attend to its own latent representations. This design fosters the emergence of working memory within the Transformer, allowing it to process indefinitely long sequences. TransformerFAM requires no additional weights, enabling seamless integration with pre-trained models. Our experiments show that TransformerFAM significantly improves Transformer performance on long-context tasks across various model sizes (1B, 8B, and 24B). These results showcase the potential to empower Large Language Models (LLMs) to process sequences of unlimited length.
Authors: Jing-Cheng Pang, Si-Hang Yang, Kaiyuan Li, Jiaji Zhang, Xiong-Hui Chen, Nan Tang, Yang Yu
Abstract: Reinforcement learning (RL) trains agents to accomplish complex tasks through environmental interaction data, but its capacity is also limited by the scope of the available data. To obtain a knowledgeable agent, a promising approach is to leverage the knowledge from large language models (LLMs). Despite previous studies combining LLMs with RL, seamless integration of the two components remains challenging due to their semantic gap. This paper introduces a novel method, Knowledgeable Agents from Language Model Rollouts (KALM), which extracts knowledge from LLMs in the form of imaginary rollouts that can be easily learned by the agent through offline reinforcement learning methods. The primary challenge of KALM lies in LLM grounding, as LLMs are inherently limited to textual data, whereas environmental data often comprise numerical vectors unseen to LLMs. To address this, KALM fine-tunes the LLM to perform various tasks based on environmental data, including bidirectional translation between natural language descriptions of skills and their corresponding rollout data. This grounding process enhances the LLM's comprehension of environmental dynamics, enabling it to generate diverse and meaningful imaginary rollouts that reflect novel skills. Initial empirical evaluations on the CLEVR-Robot environment demonstrate that KALM enables agents to complete complex rephrasings of task goals and extend their capabilities to novel tasks requiring unprecedented optimal behaviors. KALM achieves a success rate of 46% in executing tasks with unseen goals, substantially surpassing the 26% success rate achieved by baseline methods. Furthermore, KALM effectively enables the LLM to comprehend environmental dynamics, resulting in the generation of meaningful imaginary rollouts that reflect novel skills and demonstrate the seamless integration of large language models and reinforcement learning.
Authors: Mohamad Nabeel, Doumitrou Daniil Nimara, Tahar Zanouda
Abstract: In recent years, the evolution of Telecom towards achieving intelligent, autonomous, and open networks has led to an increasingly complex Telecom Software system, supporting various heterogeneous deployment scenarios, with multi-standard and multi-vendor support. As a result, it becomes a challenge for large-scale Telecom software companies to develop and test software for all deployment scenarios. To address these challenges, we propose a framework for Automated Test Generation for large-scale Telecom Software systems. We begin by generating Test Case Input data for test scenarios observed using a time-series Generative model trained on historical Telecom Network data during field trials. Additionally, the time-series Generative model helps in preserving the privacy of Telecom data. The generated time-series software performance data are then utilized with test descriptions written in natural language to generate Test Script using the Generative Large Language Model. Our comprehensive experiments on public datasets and Telecom datasets obtained from operational Telecom Networks demonstrate that the framework can effectively generate comprehensive test case data input and useful test code.
Authors: Quang Minh Dinh, Minh Khoi Ho, Anh Quan Dang, Hung Phong Tran
Abstract: Traffic video description and analysis have received much attention recently due to the growing demand for efficient and reliable urban surveillance systems. Most existing methods only focus on locating traffic event segments, which severely lack descriptive details related to the behaviour and context of all the subjects of interest in the events. In this paper, we present TrafficVLM, a novel multi-modal dense video captioning model for vehicle ego camera view. TrafficVLM models traffic video events at different levels of analysis, both spatially and temporally, and generates long fine-grained descriptions for the vehicle and pedestrian at different phases of the event. We also propose a conditional component for TrafficVLM to control the generation outputs and a multi-task fine-tuning paradigm to enhance TrafficVLM's learning capability. Experiments show that TrafficVLM performs well on both vehicle and overhead camera views. Our solution achieved outstanding results in Track 2 of the AI City Challenge 2024, ranking us third in the challenge standings. Our code is publicly available at https://github.com/quangminhdinh/TrafficVLM.
Authors: Vishwas Sathish, Hannah Lin, Aditya K Kamath, Anish Nyayachavadi
Abstract: Large Language Models (LLMs) are a powerful technology that augment human skill to create new opportunities, akin to the development of steam engines and the internet. However, LLMs come with a high cost. They require significant computing resources and energy to train and serve. Inequity in their control and access has led to concentration of ownership and power to a small collection of corporations. In our study, we collect training and inference requirements for various LLMs. We then analyze the economic strengths of nations and organizations in the context of developing and serving these models. Additionally, we also look at whether individuals around the world can access and use this emerging technology. We compare and contrast these groups to show that these technologies are monopolized by a surprisingly few entities. We conclude with a qualitative study on the ethical implications of our findings and discuss future directions towards equity in LLM access.
Authors: Karim Benharrak, Tim Zindulka, Daniel Buschek
Abstract: Large Language Models have become an integral part of new intelligent and interactive writing assistants. Many are offered commercially with a chatbot-like UI, such as ChatGPT, and provide little information about their inner workings. This makes this new type of widespread system a potential target for deceptive design patterns. For example, such assistants might exploit hidden costs by providing guidance up until a certain point before asking for a fee to see the rest. As another example, they might sneak unwanted content/edits into longer generated or revised text pieces (e.g. to influence the expressed opinion). With these and other examples, we conceptually transfer several deceptive patterns from the literature to the new context of AI writing assistants. Our goal is to raise awareness and encourage future research into how the UI and interaction design of such systems can impact people and their writing.
Authors: V\'ictor A. Braberman, Flavia Bonomo-Braberman, Yiannis Charalambous, Juan G. Colonna, Lucas C. Cordeiro, Rosiane de Freitas
Abstract: Prompting has become one of the main approaches to leverage emergent capabilities of Large Language Models [Brown et al. NeurIPS 2020, Wei et al. TMLR 2022, Wei et al. NeurIPS 2022]. During the last year, researchers and practitioners have been playing with prompts to see how to make the most of LLMs. By homogeneously dissecting 80 papers, we investigate in deep how software testing and verification research communities have been abstractly architecting their LLM-enabled solutions. More precisely, first, we want to validate whether downstream tasks are an adequate concept to convey the blueprint of prompt-based solutions. We also aim at identifying number and nature of such tasks in solutions. For such goal, we develop a novel downstream task taxonomy that enables pinpointing some engineering patterns in a rather varied spectrum of Software Engineering problems that encompasses testing, fuzzing, debugging, vulnerability detection, static analysis and program verification approaches.
Authors: Shu-wen Yang, Heng-Jui Chang, Zili Huang, Andy T. Liu, Cheng-I Lai, Haibin Wu, Jiatong Shi, Xuankai Chang, Hsiang-Sheng Tsai, Wen-Chin Huang, Tzu-hsun Feng, Po-Han Chi, Yist Y. Lin, Yung-Sung Chuang, Tzu-Hsien Huang, Wei-Cheng Tseng, Kushal Lakhotia, Shang-Wen Li, Abdelrahman Mohamed, Shinji Watanabe, Hung-yi Lee
Abstract: The foundation model paradigm leverages a shared foundation model to achieve state-of-the-art (SOTA) performance for various tasks, requiring minimal downstream-specific modeling and data annotation. This approach has proven crucial in the field of Natural Language Processing (NLP). However, the speech processing community lacks a similar setup to explore the paradigm systematically. In this work, we establish the Speech processing Universal PERformance Benchmark (SUPERB) to study the effectiveness of the paradigm for speech. We propose a unified multi-tasking framework to address speech processing tasks in SUPERB using a frozen foundation model followed by task-specialized, lightweight prediction heads. Combining our results with community submissions, we verify that the foundation model paradigm is promising for speech, and our multi-tasking framework is simple yet effective, as the best-performing foundation model shows competitive generalizability across most SUPERB tasks. For reproducibility and extensibility, we have developed a long-term maintained platform that enables deterministic benchmarking, allows for result sharing via an online leaderboard, and promotes collaboration through a community-driven benchmark database to support new development cycles. Finally, we conduct a series of analyses to offer an in-depth understanding of SUPERB and speech foundation models, including information flows across tasks inside the models, the correctness of the weighted-sum benchmarking protocol and the statistical significance and robustness of the benchmark.
Authors: Xiao Wang, Shiao Wang, Yuhe Ding, Yuehang Li, Wentao Wu, Yao Rong, Weizhe Kong, Ju Huang, Shihao Li, Haoxiang Yang, Ziwen Wang, Bo Jiang, Chenglong Li, Yaowei Wang, Yonghong Tian, Jin Tang
Abstract: In the post-deep learning era, the Transformer architecture has demonstrated its powerful performance across pre-trained big models and various downstream tasks. However, the enormous computational demands of this architecture have deterred many researchers. To further reduce the complexity of attention models, numerous efforts have been made to design more efficient methods. Among them, the State Space Model (SSM), as a possible replacement for the self-attention based Transformer model, has drawn more and more attention in recent years. In this paper, we give the first comprehensive review of these works and also provide experimental comparisons and analysis to better demonstrate the features and advantages of SSM. Specifically, we first give a detailed description of principles to help the readers quickly capture the key ideas of SSM. After that, we dive into the reviews of existing SSMs and their various applications, including natural language processing, computer vision, graph, multi-modal and multi-media, point cloud/event stream, time series data, and other domains. In addition, we give statistical comparisons and analysis of these models and hope it helps the readers to understand the effectiveness of different structures on various tasks. Then, we propose possible research points in this direction to better promote the development of the theoretical model and application of SSM. More related works will be continuously updated on the following GitHub: https://github.com/Event-AHU/Mamba_State_Space_Model_Paper_List.
URLs: https://github.com/Event-AHU/Mamba_State_Space_Model_Paper_List.
Authors: Siyan Zhao, Daniel Israel, Guy Van den Broeck, Aditya Grover
Abstract: During inference for transformer-based large language models (LLM), prefilling is the computation of the key-value (KV) cache for input tokens in the prompt prior to autoregressive generation. For longer input prompt lengths, prefilling will incur a significant overhead on decoding time. In this work, we highlight the following pitfall of prefilling: for batches containing high-varying prompt lengths, significant computation is wasted by the standard practice of padding sequences to the maximum length. As LLMs increasingly support longer context lengths, potentially up to 10 million tokens, variations in prompt lengths within a batch become more pronounced. To address this, we propose Prepacking, a simple yet effective method to optimize prefilling computation. To avoid redundant computation on pad tokens, prepacking combines prompts of varying lengths into a sequence and packs multiple sequences into a compact batch using a bin-packing algorithm. It then modifies the attention mask and positional encoding to compute multiple prefilled KV-caches for multiple prompts within a single sequence. On standard curated dataset containing prompts with varying lengths, we obtain a significant speed and memory efficiency improvements as compared to the default padding-based prefilling computation within Huggingface across a range of base model configurations and inference serving scenarios.
Authors: Taichi Sakaguchi, Akira Taniguchi, Yoshinobu Hagiwara, Lotfi El Hafi, Shoichi Hasegawa, Tadahiro Taniguchi
Abstract: Improving instance-specific image goal navigation (InstanceImageNav), which locates the identical object in a real-world environment from a query image, is essential for robotic systems to assist users in finding desired objects. The challenge lies in the domain gap between low-quality images observed by the moving robot, characterized by motion blur and low-resolution, and high-quality query images provided by the user. Such domain gaps could significantly reduce the task success rate but have not been the focus of previous work. To address this, we propose a novel method called Few-shot Cross-quality Instance-aware Adaptation (CrossIA), which employs contrastive learning with an instance classifier to align features between massive low- and few high-quality images. This approach effectively reduces the domain gap by bringing the latent representations of cross-quality images closer on an instance basis. Additionally, the system integrates an object image collection with a pre-trained deblurring model to enhance the observed image quality. Our method fine-tunes the SimSiam model, pre-trained on ImageNet, using CrossIA. We evaluated our method's effectiveness through an InstanceImageNav task with 20 different types of instances, where the robot identifies the same instance in a real-world environment as a high-quality query image. Our experiments showed that our method improves the task success rate by up to three times compared to the baseline, a conventional approach based on SuperGlue. These findings highlight the potential of leveraging contrastive learning and image enhancement techniques to bridge the domain gap and improve object localization in robotic applications. The project website is https://emergentsystemlabstudent.github.io/DomainBridgingNav/.
URLs: https://emergentsystemlabstudent.github.io/DomainBridgingNav/.
Authors: Alexey Gorbatovski, Boris Shaposhnikov, Alexey Malakhov, Nikita Surnachev, Yaroslav Aksenov, Ian Maksimov, Nikita Balagansky, Daniil Gavrilov
Abstract: The complexity of the alignment problem stems from the fact that existing methods are unstable. Researchers continuously invent various tricks to address this shortcoming. For instance, in the fundamental Reinforcement Learning From Human Feedback (RLHF) technique of Language Model alignment, in addition to reward maximization, the Kullback-Leibler divergence between the trainable policy and the SFT policy is minimized. This addition prevents the model from being overfitted to the Reward Model (RM) and generating texts that are out-of-domain for the RM. The Direct Preference Optimization (DPO) method reformulates the optimization task of RLHF and eliminates the Reward Model while tacitly maintaining the requirement for the policy to be close to the SFT policy. In our paper, we argue that this implicit limitation in the DPO method leads to sub-optimal results. We propose a new method called Trust Region DPO (TR-DPO), which updates the reference policy during training. With such a straightforward update, we demonstrate the effectiveness of TR-DPO against DPO on the Anthropic HH and TLDR datasets. We show that TR-DPO outperforms DPO by up to 19%, measured by automatic evaluation with GPT-4. The new alignment approach that we propose allows us to improve the quality of models across several parameters at once, such as coherence, correctness, level of detail, helpfulness, and harmlessness.
Authors: Chenwei Lin, Hanjia Lyu, Jiebo Luo, Xian Xu
Abstract: The emergence of Large Multimodal Models (LMMs) marks a significant milestone in the development of artificial intelligence. Insurance, as a vast and complex discipline, involves a wide variety of data forms in its operational processes, including text, images, and videos, thereby giving rise to diverse multimodal tasks. Despite this, there has been limited systematic exploration of multimodal tasks specific to insurance, nor a thorough investigation into how LMMs can address these challenges. In this paper, we explore GPT-4V's capabilities in the insurance domain. We categorize multimodal tasks by focusing primarily on visual aspects based on types of insurance (e.g., auto, household/commercial property, health, and agricultural insurance) and insurance stages (e.g., risk assessment, risk monitoring, and claims processing). Our experiment reveals that GPT-4V exhibits remarkable abilities in insurance-related tasks, demonstrating not only a robust understanding of multimodal content in the insurance domain but also a comprehensive knowledge of insurance scenarios. However, there are notable shortcomings: GPT-4V struggles with detailed risk rating and loss assessment, suffers from hallucination in image understanding, and shows variable support for different languages. Through this work, we aim to bridge the insurance domain with cutting-edge LMM technology, facilitate interdisciplinary exchange and development, and provide a foundation for the continued advancement and evolution of future research endeavors.
Authors: Guangyan Li, Yongqiang Tang, Wensheng Zhang
Abstract: Large language models (LLMs) show excellent performance in difficult tasks, but they often require massive memories and computational resources. How to reduce the parameter scale of LLMs has become research hotspots. In this study, we make an important observation that the multi-head self-attention (MHA) sub-layer of Transformer exhibits noticeable low-rank structure, while the feed-forward network (FFN) sub-layer does not. With this regard, we design a mixed compression model, which organically combines Low-Rank matrix approximation And structured Pruning (LoRAP). For the MHA sub-layer, we propose an input activation weighted singular value decomposition method to strengthen the low-rank characteristic. Furthermore, we discover that the weight matrices in MHA sub-layer have different low-rank degrees. Thus, a novel parameter allocation scheme according to the discrepancy of low-rank degrees is devised. For the FFN sub-layer, we propose a gradient-free structured channel pruning method. During the pruning, we get an interesting finding that the least important 1% of parameter actually play a vital role in model performance. Extensive evaluations on zero-shot perplexity and zero-shot task classification indicate that our proposal is superior to previous structured compression rivals under multiple compression ratios.
Authors: Daniil Merkulov, Daria Cherniuk, Alexander Rudikov, Ivan Oseledets, Ekaterina Muravleva, Aleksandr Mikhalev, Boris Kashin
Abstract: In this paper, we introduce an algorithm for data quantization based on the principles of Kashin representation. This approach hinges on decomposing any given vector, matrix, or tensor into two factors. The first factor maintains a small infinity norm, while the second exhibits a similarly constrained norm when multiplied by an orthogonal matrix. Surprisingly, the entries of factors after decomposition are well-concentrated around several peaks, which allows us to efficiently replace them with corresponding centroids for quantization purposes. We study the theoretical properties of the proposed approach and rigorously evaluate our compression algorithm in the context of next-word prediction tasks and on a set of downstream tasks for text classification. Our findings demonstrate that Kashin Quantization achieves competitive or superior quality in model performance while ensuring data compression, marking a significant advancement in the field of data quantization.
Authors: Francis McCann Ramirez, Luka Chkhetiani, Andrew Ehrenberg, Robert McHardy, Rami Botros, Yash Khare, Andrea Vanzo, Taufiquzzaman Peyash, Gabriel Oexle, Michael Liang, Ilya Sklyar, Enver Fakhan, Ahmed Efty, Daniel McCrystal, Sam Flamini, Domenic Donato, Takuya Yoshioka
Abstract: This paper describes AssemblyAI's industrial-scale automatic speech recognition (ASR) system, designed to meet the requirements of large-scale, multilingual ASR serving various application needs. Our system leverages a diverse training dataset comprising unsupervised (12.5M hours), supervised (188k hours), and pseudo-labeled (1.6M hours) data across four languages. We provide a detailed description of our model architecture, consisting of a full-context 600M-parameter Conformer encoder pre-trained with BEST-RQ and an RNN-T decoder fine-tuned jointly with the encoder. Our extensive evaluation demonstrates competitive word error rates (WERs) against larger and more computationally expensive models, such as Whisper large and Canary-1B. Furthermore, our architectural choices yield several key advantages, including an improved code-switching capability, a 5x inference speedup compared to an optimized Whisper baseline, a 30% reduction in hallucination rate on speech data, and a 90% reduction in ambient noise compared to Whisper, along with significantly improved time-stamp accuracy. Throughout this work, we adopt a system-centric approach to analyzing various aspects of fully-fledged ASR models to gain practically relevant insights useful for real-world services operating at scale.
Authors: Rohan Padhye
Abstract: The recent proliferation of generative artificial intelligence (GenAI) technologies such as pre-trained large language models (LLMs) has opened up new frontiers in computational law. An exciting area of development is the use of AI to automate the rule-based reasoning inherent in statutory and contract law. While this form of reasoning has long been studied using classical techniques of natural language processing (NLP) and formal logic, recent solutions increasingly make use of LLMs; though they are far from perfect. The advent of GenAI has made it possible to treat many of these natural language documents essentially as programs that compute a result given some set of facts. As such, it should be possible to understand, debug, maintain, evolve, and fix these documents using well-studied techniques from the field of software engineering. This article introduces several concepts of automated software testing and program analysis that could potentially be useful in computational law when applied to AI-driven analysis of statutes and contracts.
Authors: Peter Baile Chen, Yi Zhang, Dan Roth
Abstract: Retrieving relevant tables containing the necessary information to accurately answer a given question over tables is critical to open-domain question-answering (QA) systems. Previous methods assume the answer to such a question can be found either in a single table or multiple tables identified through question decomposition or rewriting. However, neither of these approaches is sufficient, as many questions require retrieving multiple tables and joining them through a join plan that cannot be discerned from the user query itself. If the join plan is not considered in the retrieval stage, the subsequent steps of reasoning and answering based on those retrieved tables are likely to be incorrect. To address this problem, we introduce a method that uncovers useful join relations for any query and database during table retrieval. We use a novel re-ranking method formulated as a mixed-integer program that considers not only table-query relevance but also table-table relevance that requires inferring join relationships. Our method outperforms the state-of-the-art approaches for table retrieval by up to 9.3% in F1 score and for end-to-end QA by up to 5.4% in accuracy.
Authors: Jiayi Li, Ruilin Luo, Jiaqi Sun, Jing Xiao, Yujiu Yang
Abstract: Knowledge Graph Completion (KGC) has emerged as a promising solution to address the issue of incompleteness within Knowledge Graphs (KGs). Traditional KGC research primarily centers on triple classification and link prediction. Nevertheless, we contend that these tasks do not align well with real-world scenarios and merely serve as surrogate benchmarks. In this paper, we investigate three crucial processes relevant to real-world construction scenarios: (a) the verification process, which arises from the necessity and limitations of human verifiers; (b) the mining process, which identifies the most promising candidates for verification; and (c) the training process, which harnesses verified data for subsequent utilization; in order to achieve a transition toward more realistic challenges. By integrating these three processes, we introduce the Progressive Knowledge Graph Completion (PKGC) task, which simulates the gradual completion of KGs in real-world scenarios. Furthermore, to expedite PKGC processing, we propose two acceleration modules: Optimized Top-$k$ algorithm and Semantic Validity Filter. These modules significantly enhance the efficiency of the mining procedure. Our experiments demonstrate that performance in link prediction does not accurately reflect performance in PKGC. A more in-depth analysis reveals the key factors influencing the results and provides potential directions for future research.
Authors: Usman Anwar, Abulhair Saparov, Javier Rando, Daniel Paleka, Miles Turpin, Peter Hase, Ekdeep Singh Lubana, Erik Jenner, Stephen Casper, Oliver Sourbut, Benjamin L. Edelman, Zhaowei Zhang, Mario G\"unther, Anton Korinek, Jose Hernandez-Orallo, Lewis Hammond, Eric Bigelow, Alexander Pan, Lauro Langosco, Tomasz Korbak, Heidi Zhang, Ruiqi Zhong, Se\'an \'O h\'Eigeartaigh, Gabriel Recchia, Giulio Corsi, Alan Chan, Markus Anderljung, Lilian Edwards, Yoshua Bengio, Danqi Chen, Samuel Albanie, Tegan Maharaj, Jakob Foerster, Florian Tramer, He He, Atoosa Kasirzadeh, Yejin Choi, David Krueger
Abstract: This work identifies 18 foundational challenges in assuring the alignment and safety of large language models (LLMs). These challenges are organized into three different categories: scientific understanding of LLMs, development and deployment methods, and sociotechnical challenges. Based on the identified challenges, we pose $200+$ concrete research questions.
Authors: Navonil Majumder, Chia-Yu Hung, Deepanway Ghosal, Wei-Ning Hsu, Rada Mihalcea, Soujanya Poria
Abstract: Generative multimodal content is increasingly prevalent in much of the content creation arena, as it has the potential to allow artists and media personnel to create pre-production mockups by quickly bringing their ideas to life. The generation of audio from text prompts is an important aspect of such processes in the music and film industry. Many of the recent diffusion-based text-to-audio models focus on training increasingly sophisticated diffusion models on a large set of datasets of prompt-audio pairs. These models do not explicitly focus on the presence of concepts or events and their temporal ordering in the output audio with respect to the input prompt. Our hypothesis is focusing on how these aspects of audio generation could improve audio generation performance in the presence of limited data. As such, in this work, using an existing text-to-audio model Tango, we synthetically create a preference dataset where each prompt has a winner audio output and some loser audio outputs for the diffusion model to learn from. The loser outputs, in theory, have some concepts from the prompt missing or in an incorrect order. We fine-tune the publicly available Tango text-to-audio model using diffusion-DPO (direct preference optimization) loss on our preference dataset and show that it leads to improved audio output over Tango and AudioLDM2, in terms of both automatic- and manual-evaluation metrics.
Authors: Ziniu Zhang, Shulin Tian, Liangyu Chen, Ziwei Liu
Abstract: Autonomous embodied agents live on an Internet of multimedia websites. Can they hop around multimodal websites to complete complex user tasks? Existing benchmarks fail to assess them in a realistic, evolving environment for their embodiment across websites. To answer this question, we present MMInA, a multihop and multimodal benchmark to evaluate the embodied agents for compositional Internet tasks, with several appealing properties: 1) Evolving real-world multimodal websites. Our benchmark uniquely operates on evolving real-world websites, ensuring a high degree of realism and applicability to natural user tasks. Our data includes 1,050 human-written tasks covering various domains such as shopping and travel, with each task requiring the agent to autonomously extract multimodal information from web pages as observations; 2) Multihop web browsing. Our dataset features naturally compositional tasks that require information from or actions on multiple websites to solve, to assess long-range reasoning capabilities on web tasks; 3) Holistic evaluation. We propose a novel protocol for evaluating an agent's progress in completing multihop tasks. We experiment with both standalone (multimodal) language models and heuristic-based web agents. Extensive experiments demonstrate that while long-chain multihop web tasks are easy for humans, they remain challenging for state-of-the-art web agents. We identify that agents are more likely to fail on the early hops when solving tasks of more hops, which results in lower task success rates. To address this issue, we propose a simple memory augmentation approach replaying past action trajectories to reflect. Our method significantly improved both the single-hop and multihop web browsing abilities of agents. See our code and data at https://mmina.cliangyu.com
Authors: Chang Tian, Wenpeng Yin, Marie-Francine Moens
Abstract: A dialogue policy module is an essential part of task-completion dialogue systems. Recently, increasing interest has focused on reinforcement learning (RL)-based dialogue policy. Its favorable performance and wise action decisions rely on an accurate estimation of action values. The overestimation problem is a widely known issue of RL since its estimate of the maximum action value is larger than the ground truth, which results in an unstable learning process and suboptimal policy. This problem is detrimental to RL-based dialogue policy learning. To mitigate this problem, this paper proposes a dynamic partial average estimator (DPAV) of the ground truth maximum action value. DPAV calculates the partial average between the predicted maximum action value and minimum action value, where the weights are dynamically adaptive and problem-dependent. We incorporate DPAV into a deep Q-network as the dialogue policy and show that our method can achieve better or comparable results compared to top baselines on three dialogue datasets of different domains with a lower computational load. In addition, we also theoretically prove the convergence and derive the upper and lower bounds of the bias compared with those of other methods.
Authors: Yunhua Zhou, Pengyu Wang, Peiju Liu, Yuxin Wang, Xipeng Qiu
Abstract: Most existing methods of Out-of-Domain (OOD) intent classification rely on extensive auxiliary OOD corpora or specific training paradigms. However, they are underdeveloped in the underlying principle that the models should have differentiated confidence in In- and Out-of-domain intent. In this work, we shed light on the fundamental cause of model overconfidence on OOD and demonstrate that calibrated subnetworks can be uncovered by pruning the overparameterized model. Calibrated confidence provided by the subnetwork can better distinguish In- and Out-of-domain, which can be a benefit for almost all post hoc methods. In addition to bringing fundamental insights, we also extend the Lottery Ticket Hypothesis to open-world scenarios. We conduct extensive experiments on four real-world datasets to demonstrate our approach can establish consistent improvements compared with a suite of competitive baselines.
Authors: Goro Kobayashi, Tatsuki Kuribayashi, Sho Yokoi, Kentaro Inui
Abstract: Transformers are ubiquitous in wide tasks. Interpreting their internals is a pivotal goal. Nevertheless, their particular components, feed-forward (FF) blocks, have typically been less analyzed despite their substantial parameter amounts. We analyze the input contextualization effects of FF blocks by rendering them in the attention maps as a human-friendly visualization scheme. Our experiments with both masked- and causal-language models reveal that FF networks modify the input contextualization to emphasize specific types of linguistic compositions. In addition, FF and its surrounding components tend to cancel out each other's effects, suggesting potential redundancy in the processing of the Transformer layer.
Authors: Ning Lu, Shengcai Liu, Zhirui Zhang, Qi Wang, Haifeng Liu, Ke Tang
Abstract: Word-level textual adversarial attacks have demonstrated notable efficacy in misleading Natural Language Processing (NLP) models. Despite their success, the underlying reasons for their effectiveness and the fundamental characteristics of adversarial examples (AEs) remain obscure. This work aims to interpret word-level attacks by examining their $n$-gram frequency patterns. Our comprehensive experiments reveal that in approximately 90\% of cases, word-level attacks lead to the generation of examples where the frequency of $n$-grams decreases, a tendency we term as the $n$-gram Frequency Descend ($n$-FD). This finding suggests a straightforward strategy to enhance model robustness: training models using examples with $n$-FD. To examine the feasibility of this strategy, we employed the $n$-gram frequency information, as an alternative to conventional loss gradients, to generate perturbed examples in adversarial training. The experiment results indicate that the frequency-based approach performs comparably with the gradient-based approach in improving model robustness. Our research offers a novel and more intuitive perspective for understanding word-level textual adversarial attacks and proposes a new direction to improve model robustness.
Authors: Ruohong Zhang, Yau-Shian Wang, Yiming Yang
Abstract: The remarkable performance of large language models (LLMs) in zero-shot language understanding has garnered significant attention. However, employing LLMs for large-scale inference or domain-specific fine-tuning requires immense computational resources due to their substantial model size. To overcome these limitations, we introduce a novel method, namely GenCo, which leverages the strong generative power of LLMs to assist in training a smaller and more adaptable language model. In our method, an LLM plays an important role in the self-training loop of a smaller model in two important ways. Firstly, the LLM is used to augment each input instance with a variety of possible continuations, enriching its semantic context for better understanding. Secondly, it helps crafting additional high-quality training pairs, by rewriting input texts conditioned on predicted labels. This ensures the generated texts are highly relevant to the predicted labels, alleviating the prediction error during pseudo-labeling, while reducing the dependency on large volumes of unlabeled text. In our experiments, GenCo outperforms previous state-of-the-art methods when only limited ($<5\%$ of original) in-domain text data is available. Notably, our approach surpasses the performance of Alpaca-7B with human prompts, highlighting the potential of leveraging LLM for self-training.
Authors: Qiang Liu, Junfei Wu, Shu Wu, Liang Wang
Abstract: Evidence-aware fake news detection aims to conduct reasoning between news and evidence, which is retrieved based on news content, to find uniformity or inconsistency. However, we find evidence-aware detection models suffer from biases, i.e., spurious correlations between news/evidence contents and true/fake news labels, and are hard to be generalized to Out-Of-Distribution (OOD) situations. To deal with this, we propose a novel Dual Adversarial Learning (DAL) approach. We incorporate news-aspect and evidence-aspect debiasing discriminators, whose targets are both true/fake news labels, in DAL. Then, DAL reversely optimizes news-aspect and evidence-aspect debiasing discriminators to mitigate the impact of news and evidence content biases. At the same time, DAL also optimizes the main fake news predictor, so that the news-evidence interaction module can be learned. This process allows us to teach evidence-aware fake news detection models to better conduct news-evidence reasoning, and minimize the impact of content biases. To be noted, our proposed DAL approach is a plug-and-play module that works well with existing backbones. We conduct comprehensive experiments under two OOD settings, and plug DAL in four evidence-aware fake news detection backbones. Results demonstrate that, DAL significantly and stably outperforms the original backbones and some competitive debiasing methods.
Authors: Xingyu Fu, Ben Zhou, Sihao Chen, Mark Yatskar, Dan Roth
Abstract: Recent advances in multimodal large language models (LLMs) have shown extreme effectiveness in visual question answering (VQA). However, the design nature of these end-to-end models prevents them from being interpretable to humans, undermining trust and applicability in critical domains. While post-hoc rationales offer certain insight into understanding model behavior, these explanations are not guaranteed to be faithful to the model. In this paper, we address these shortcomings by introducing an interpretable by design model that factors model decisions into intermediate human-legible explanations, and allows people to easily understand why a model fails or succeeds. We propose the Dynamic Clue Bottleneck Model ( (DCLUB), a method that is designed towards an inherently interpretable VQA system. DCLUB provides an explainable intermediate space before the VQA decision and is faithful from the beginning, while maintaining comparable performance to black-box systems. Given a question, DCLUB first returns a set of visual clues: natural language statements of visually salient evidence from the image, and then generates the output based solely on the visual clues. To supervise and evaluate the generation of VQA explanations within DCLUB, we collect a dataset of 1.7k reasoning-focused questions with visual clues. Evaluations show that our inherently interpretable system can improve 4.64% over a comparable black-box system in reasoning-focused questions while preserving 99.43% of performance on VQA-v2.
Authors: Jiahuan Li, Hao Zhou, Shujian Huang, Shanbo Cheng, Jiajun Chen
Abstract: Large-scale Pretrained Language Models (LLMs), such as ChatGPT and GPT4, have shown strong abilities in multilingual translations, without being explicitly trained on parallel corpora. It is interesting how the LLMs obtain their ability to carry out translation instructions for different languages. In this paper, we present a detailed analysis by finetuning a multilingual pretrained language model, XGLM-7B, to perform multilingual translation following given instructions. Firstly, we show that multilingual LLMs have stronger translation abilities than previously demonstrated. For a certain language, the performance depends on its similarity to English and the amount of data used in the pretraining phase. Secondly, we find that LLMs' ability to carry out translation instructions relies on the understanding of translation instructions and the alignment among different languages. With multilingual finetuning, LLMs could learn to perform the translation task well even for those language pairs unseen during the instruction tuning phase.
Authors: Fuxiao Liu, Paiheng Xu, Zongxia Li, Yue Feng
Abstract: We investigate the role of various demonstration components in the in-context learning (ICL) performance of large language models (LLMs). Specifically, we explore the impacts of ground-truth labels, input distribution, and complementary explanations, particularly when these are altered or perturbed. We build on previous work, which offers mixed findings on how these elements influence ICL. To probe these questions, we employ explainable NLP (XNLP) methods and utilize saliency maps of contrastive demonstrations for both qualitative and quantitative analysis. Our findings reveal that flipping ground-truth labels significantly affects the saliency, though it's more noticeable in larger LLMs. Our analysis of the input distribution at a granular level reveals that changing sentiment-indicative terms in a sentiment analysis task to neutral ones does not have as substantial an impact as altering ground-truth labels. Finally, we find that the effectiveness of complementary explanations in boosting ICL performance is task-dependent, with limited benefits seen in sentiment analysis tasks compared to symbolic reasoning tasks. These insights are critical for understanding the functionality of LLMs and guiding the development of effective demonstrations, which is increasingly relevant in light of the growing use of LLMs in applications such as ChatGPT. Our research code is publicly available at https://github.com/paihengxu/XICL.
Authors: Seonghyeon Ye, Doyoung Kim, Sungdong Kim, Hyeonbin Hwang, Seungone Kim, Yongrae Jo, James Thorne, Juho Kim, Minjoon Seo
Abstract: Evaluation of Large Language Models (LLMs) is challenging because instruction-following necessitates alignment with human values and the required set of skills varies depending on the instruction. However, previous studies have mainly focused on coarse-grained evaluation (i.e. overall preference-based evaluation), which limits interpretability since it does not consider the nature of user instructions that require instance-wise skill composition. In this paper, we introduce FLASK (Fine-grained Language Model Evaluation based on Alignment Skill Sets), a fine-grained evaluation protocol for both human-based and model-based evaluation which decomposes coarse-level scoring to a skill set-level scoring for each instruction. We experimentally observe that the fine-graininess of evaluation is crucial for attaining a holistic view of model performance and increasing the reliability of the evaluation. Using FLASK, we compare multiple open-source and proprietary LLMs and observe a high correlation between model-based and human-based evaluations. We publicly release the evaluation data and code implementation at https://github.com/kaistAI/FLASK.
Authors: Flor Miriam Plaza-del-Arco, Debora Nozza, Dirk Hovy
Abstract: Large Language Models (LLMs) exhibit remarkable text classification capabilities, excelling in zero- and few-shot learning (ZSL and FSL) scenarios. However, since they are trained on different datasets, performance varies widely across tasks between those models. Recent studies emphasize the importance of considering human label variation in data annotation. However, how this human label variation also applies to LLMs remains unexplored. Given this likely model specialization, we ask: Do aggregate LLM labels improve over individual models (as for human annotators)? We evaluate four recent instruction-tuned LLMs as annotators on five subjective tasks across four languages. We use ZSL and FSL setups and label aggregation from human annotation. Aggregations are indeed substantially better than any individual model, benefiting from specialization in diverse tasks or languages. Surprisingly, FSL does not surpass ZSL, as it depends on the quality of the selected examples. However, there seems to be no good information-theoretical strategy to select those. We find that no LLM method rivals even simple supervised models. We also discuss the tradeoffs in accuracy, cost, and moral/ethical considerations between LLM and human annotation.
Authors: Jiang Li, Xiaoping Wang, Yingjian Liu, Zhigang Zeng
Abstract: Multimodal emotion recognition in conversation (ERC) has garnered growing attention from research communities in various fields. In this paper, we propose a Cross-modal Fusion Network with Emotion-Shift Awareness (CFN-ESA) for ERC. Extant approaches employ each modality equally without distinguishing the amount of emotional information in these modalities, rendering it hard to adequately extract complementary information from multimodal data. To cope with this problem, in CFN-ESA, we treat textual modality as the primary source of emotional information, while visual and acoustic modalities are taken as the secondary sources. Besides, most multimodal ERC models ignore emotion-shift information and overfocus on contextual information, leading to the failure of emotion recognition under emotion-shift scenario. We elaborate an emotion-shift module to address this challenge. CFN-ESA mainly consists of unimodal encoder (RUME), cross-modal encoder (ACME), and emotion-shift module (LESM). RUME is applied to extract conversation-level contextual emotional cues while pulling together data distributions between modalities; ACME is utilized to perform multimodal interaction centered on textual modality; LESM is used to model emotion shift and capture emotion-shift information, thereby guiding the learning of the main task. Experimental results demonstrate that CFN-ESA can effectively promote performance for ERC and remarkably outperform state-of-the-art models.
Authors: Lang Cao
Abstract: A significant application of Large Language Models (LLMs), like ChatGPT, is their deployment as chat agents, which respond to human inquiries across a variety of domains. While current LLMs proficiently answer general questions, they often fall short in complex diagnostic scenarios such as legal, medical, or other specialized consultations. These scenarios typically require Task-Oriented Dialogue (TOD), where an AI chat agent must proactively pose questions and guide users toward specific goals or task completion. Previous fine-tuning models have underperformed in TOD and the full potential of conversational capability in current LLMs has not yet been fully explored. In this paper, we introduce DiagGPT (Dialogue in Diagnosis GPT), an innovative approach that extends LLMs to more TOD scenarios. In addition to guiding users to complete tasks, DiagGPT can effectively manage the status of all topics throughout the dialogue development. This feature enhances user experience and offers a more flexible interaction in TOD. Our experiments demonstrate that DiagGPT exhibits outstanding performance in conducting TOD with users, showing its potential for practical applications in various fields.
Authors: Ahmed Abdeen Hamed, Xindong Wu
Abstract: Generative AI tools exemplified by ChatGPT are becoming a new reality. This study is motivated by the premise that ``AI generated content may exhibit a distinctive behavior that can be separated from scientific articles''. In this study, we show how articles can be generated using means of prompt engineering for various diseases and conditions. We then show how we tested this premise in two phases and prove its validity. Subsequently, we introduce xFakeSci, a novel learning algorithm, that is capable of distinguishing ChatGPT-generated articles from publications produced by scientists. The algorithm is trained using network models driven from both sources. As for the classification step, it was performed using 300 articles per condition. The actual label steps took place against an equal mix of 50 generated articles and 50 authentic PubMed abstracts. The testing also spanned publication periods from 2010 to 2024 and encompassed research on three distinct diseases: cancer, depression, and Alzheimer's. Further, we evaluated the accuracy of the xFakeSci algorithm against some of the classical data mining algorithms (e.g., Support Vector Machines, Regression, and Naive Bayes). The xFakeSci algorithm achieved F1 scores ranging from 80% to 94%, outperforming common data mining algorithms, which scored F1 values between 38% and 52%. We attribute the noticeable difference to the introduction of calibration and a proximity distance heuristic, which underscores this promising performance. Indeed, the prediction of fake science generated by ChatGPT presents a considerable challenge. Nonetheless, the introduction of the xFakeSci algorithm is a significant step on the way to combating fake science.
Authors: Spencer M. Seals, Valerie L. Shalin
Abstract: The development of highly fluent large language models (LLMs) has prompted increased interest in assessing their reasoning and problem-solving capabilities. We investigate whether several LLMs can solve a classic type of deductive reasoning problem from the cognitive science literature. The tested LLMs have limited abilities to solve these problems in their conventional form. We performed follow up experiments to investigate if changes to the presentation format and content improve model performance. We do find performance differences between conditions; however, they do not improve overall performance. Moreover, we find that performance interacts with presentation format and content in unexpected ways that differ from human performance. Overall, our results suggest that LLMs have unique reasoning biases that are only partially predicted from human reasoning performance and the human-generated language corpora that informs them.
Authors: Boshko Koloski, Bla\v{z} \v{S}krlj, Marko Robnik-\v{S}ikonja, Senja Pollak
Abstract: The cross-lingual transfer is a promising technique to solve tasks in less-resourced languages. In this empirical study, we compare two fine-tuning approaches combined with zero-shot and full-shot learning approaches for large language models in a cross-lingual setting. As fine-tuning strategies, we compare parameter-efficient adapter methods with fine-tuning of all parameters. As cross-lingual transfer strategies, we compare the intermediate-training (\textit{IT}) that uses each language sequentially and cross-lingual validation (\textit{CLV}) that uses a target language already in the validation phase of fine-tuning. We assess the success of transfer and the extent of catastrophic forgetting in a source language due to cross-lingual transfer, i.e., how much previously acquired knowledge is lost when we learn new information in a different language. The results on two different classification problems, hate speech detection and product reviews, each containing datasets in several languages, show that the \textit{IT} cross-lingual strategy outperforms \textit{CLV} for the target language. Our findings indicate that, in the majority of cases, the \textit{CLV} strategy demonstrates superior retention of knowledge in the base language (English) compared to the \textit{IT} strategy, when evaluating catastrophic forgetting in multiple cross-lingual transfers.
Authors: Tanmay Parekh, I-Hung Hsu, Kuan-Hao Huang, Kai-Wei Chang, Nanyun Peng
Abstract: Label projection, which involves obtaining translated labels and texts jointly, is essential for leveraging machine translation to facilitate cross-lingual transfer in structured prediction tasks. Prior research exploring label projection often compromise translation accuracy by favoring simplified label translation or relying solely on word-level alignments. In this paper, we introduce a novel label projection approach, CLaP, which translates text to the target language and performs contextual translation on the labels using the translated text as the context, ensuring better accuracy for the translated labels. We leverage instruction-tuned language models with multilingual capabilities as our contextual translator, imposing the constraint of the presence of translated labels in the translated text via instructions. We benchmark CLaP with other label projection techniques on zero-shot cross-lingual transfer across 39 languages on two representative structured prediction tasks - event argument extraction (EAE) and named entity recognition (NER), showing over 2.4 F1 improvement for EAE and 1.4 F1 improvement for NER. We further explore the applicability of CLaP on ten extremely low-resource languages to showcase its potential for cross-lingual structured prediction.
Authors: Suhas Kotha, Jacob Mitchell Springer, Aditi Raghunathan
Abstract: We lack a systematic understanding of the effects of fine-tuning (via methods such as instruction-tuning or reinforcement learning from human feedback), particularly on tasks outside the narrow fine-tuning distribution. In a simplified scenario, we demonstrate that improving performance on tasks within the fine-tuning data distribution comes at the expense of capabilities on other tasks. We hypothesize that language models implicitly infer the task of the prompt and that fine-tuning skews this inference towards tasks in the fine-tuning distribution. To test this, we propose Conjugate Prompting, which artificially makes the task look farther from the fine-tuning distribution while requiring the same capability, and we find that this recovers some of the pretraining capabilities in our synthetic setup. Since real-world fine-tuning distributions are predominantly English, we apply conjugate prompting to recover pretrained capabilities in LLMs by simply translating the prompts to different languages. This allows us to recover in-context learning abilities lost via instruction tuning, natural reasoning capability lost during code fine-tuning, and, more concerningly, harmful content generation suppressed by safety fine-tuning in chatbots like ChatGPT.
Authors: Canyu Chen, Kai Shu
Abstract: The advent of Large Language Models (LLMs) has made a transformative impact. However, the potential that LLMs such as ChatGPT can be exploited to generate misinformation has posed a serious concern to online safety and public trust. A fundamental research question is: will LLM-generated misinformation cause more harm than human-written misinformation? We propose to tackle this question from the perspective of detection difficulty. We first build a taxonomy of LLM-generated misinformation. Then we categorize and validate the potential real-world methods for generating misinformation with LLMs. Then, through extensive empirical investigation, we discover that LLM-generated misinformation can be harder to detect for humans and detectors compared to human-written misinformation with the same semantics, which suggests it can have more deceptive styles and potentially cause more harm. We also discuss the implications of our discovery on combating misinformation in the age of LLMs and the countermeasures.
Authors: Yapei Chang, Kyle Lo, Tanya Goyal, Mohit Iyyer
Abstract: Summarizing book-length documents (>100K tokens) that exceed the context window size of large language models (LLMs) requires first breaking the input document into smaller chunks and then prompting an LLM to merge, update, and compress chunk-level summaries. Despite the complexity and importance of this task, it has yet to be meaningfully studied due to the challenges of evaluation: existing book-length summarization datasets (e.g., BookSum) are in the pretraining data of most public LLMs, and existing evaluation methods struggle to capture errors made by modern LLM summarizers. In this paper, we present the first study of the coherence of LLM-based book-length summarizers implemented via two prompting workflows: (1) hierarchically merging chunk-level summaries, and (2) incrementally updating a running summary. We obtain 1193 fine-grained human annotations on GPT-4 generated summaries of 100 recently-published books and identify eight common types of coherence errors made by LLMs. Because human evaluation is expensive and time-consuming, we develop an automatic metric, BooookScore, that measures the proportion of sentences in a summary that do not contain any of the identified error types. BooookScore has high agreement with human annotations and allows us to systematically evaluate the impact of many other critical parameters (e.g., chunk size, base LLM) while saving $15K USD and 500 hours in human evaluation costs. We find that closed-source LLMs such as GPT-4 and Claude 2 produce summaries with higher BooookScore than those generated by open-source models. While LLaMA 2 falls behind other models, Mixtral achieves performance on par with GPT-3.5-Turbo. Incremental updating yields lower BooookScore but higher level of detail than hierarchical merging, a trade-off sometimes preferred by annotators.
Authors: Kuan Wang, Yadong Lu, Michael Santacroce, Yeyun Gong, Chao Zhang, Yelong Shen
Abstract: Recent advances in large language models (LLMs) have demonstrated potential for LLM agents. To facilitate the training for these agents with both linguistic feedback and non-linguistic reward signals, we introduce Learning through Communication (LTC). We design a universal buffer to store all the feedback, and an iterative pipeline to enable an LLM agent to explore and update its policy in an given environment. To optimize agent interactions for task-specific learning with our universal buffer and pipeline, we introduce diverse communication patterns tailored for both single-agent and multi-agent environments. We evaluate the efficacy of our LTC approach on four diverse datasets: ALFWorld (single-agent), HotpotQA (multi-agent collaboration), Chameleon (multi-agent competition), and GSM8k (multi-agent teacher-student). On these data sets, LTC outperforms the supervised instruction fine-tuning baselines by 3.6% to 12%. These results highlight the versatility and efficiency of LTC in facilitating online adaptation for LLM agents.
Authors: Xintong Wang, Xiaoyu Li, Xingshan Li, Chris Biemann
Abstract: Large Language Models (LLMs) have emerged as dominant foundational models in modern NLP. However, the understanding of their prediction processes and internal mechanisms, such as feed-forward networks (FFN) and multi-head self-attention (MHSA), remains largely unexplored. In this work, we probe LLMs from a human behavioral perspective, correlating values from LLMs with eye-tracking measures, which are widely recognized as meaningful indicators of human reading patterns. Our findings reveal that LLMs exhibit a similar prediction pattern with humans but distinct from that of Shallow Language Models (SLMs). Moreover, with the escalation of LLM layers from the middle layers, the correlation coefficients also increase in FFN and MHSA, indicating that the logits within FFN increasingly encapsulate word semantics suitable for predicting tokens from the vocabulary.
Authors: Yichi Zhang, Zhuo Chen, Lingbing Guo, Yajing Xu, Wen Zhang, Huajun Chen
Abstract: Large language model (LLM) based knowledge graph completion (KGC) aims to predict the missing triples in the KGs with LLMs. However, research about LLM-based KGC fails to sufficiently harness LLMs' inference proficiencies, overlooking critical structural information integral to KGs. In this paper, we explore methods to incorporate structural information into the LLMs, with the overarching goal of facilitating structure-aware reasoning. We first discuss on the existing LLM paradigms like in-context learning and instruction tuning, proposing basic structural information injection approaches. Then we propose a Knowledge Prefix Adapter (KoPA) to fulfill this stated goal. The KoPA uses a structural pre-training phase to comprehend the intricate entities and relations within KGs, representing them as structural embeddings. Then KoPA communicates such cross-modal structural information understanding to the LLMs through a knowledge prefix adapter which projects the structural embeddings into the textual space and obtains virtual knowledge tokens positioned as a prefix of the input prompt. We conduct comprehensive experiments and provide incisive analysis concerning how the introduction of cross-modal structural information would be better for LLM's factual knowledge reasoning ability. Our code and data are available at https://github.com/zjukg/KoPA .
Authors: Yutaro Yamada, Yihan Bao, Andrew K. Lampinen, Jungo Kasai, Ilker Yildirim
Abstract: Large language models (LLMs) show remarkable capabilities across a variety of tasks. Despite the models only seeing text in training, several recent studies suggest that LLM representations implicitly capture aspects of the underlying grounded concepts. Here, we explore LLM representations of a particularly salient kind of grounded knowledge -- spatial relationships. We design natural-language navigation tasks and evaluate the ability of LLMs, in particular GPT-3.5-turbo, GPT-4, and Llama2 series models, to represent and reason about spatial structures. These tasks reveal substantial variability in LLM performance across different spatial structures, including square, hexagonal, and triangular grids, rings, and trees. In extensive error analysis, we find that LLMs' mistakes reflect both spatial and non-spatial factors. These findings suggest that LLMs appear to capture certain aspects of spatial structure implicitly, but room for improvement remains.
Authors: Adri\'an Bazaga, Pietro Li\`o, Gos Micklem
Abstract: In recent years, there has been growing interest in text-to-SQL translation, which is the task of converting natural language questions into executable SQL queries. This technology is important for its potential to democratize data extraction from databases. However, some of its key hurdles include domain generalisation, which is the ability to adapt to previously unseen databases, and alignment of natural language questions with the corresponding SQL queries. To overcome these challenges, we introduce SQLformer, a novel Transformer architecture specifically crafted to perform text-to-SQL translation tasks. Our model predicts SQL queries as abstract syntax trees (ASTs) in an autoregressive way, incorporating structural inductive bias in the encoder and decoder layers. This bias, guided by database table and column selection, aids the decoder in generating SQL query ASTs represented as graphs in a Breadth-First Search canonical order. Comprehensive experiments show the state-of-the-art performance of SQLformer across five widely used text-to-SQL benchmarks. Our implementation is available at https://github.com/AdrianBZG/SQLformer.
Authors: Yixin Wan, Fanyou Wu, Weijie Xu, Srinivasan H. Sengamedu
Abstract: In this work, we propose sequence-level certainty as a common theme over hallucination in Knowledge Grounded Dialogue Generation (KGDG). We explore the correlation between the level of hallucination in model responses and two types of sequence-level certainty: probabilistic certainty and semantic certainty. Empirical results reveal that higher levels of both types of certainty in model responses are correlated with lower levels of hallucination. We further propose Certainty-based Response Ranking (CRR), a decoding-time hallucination mitigation method that samples several response candidates, ranks them based on sequence-level certainty, and outputs the response with the highest certainty level. Aligning with our definitions of sequence-level certainty, we design 2 types of CRR approaches: Probabilistic CRR (P-CRR) and Semantic CRR (S-CRR). P-CRR ranks individually sampled model responses using the arithmetic mean log-probability of the entire sequence. S-CRR approaches certainty estimation from meaning-space, and ranks model response candidates based on their semantic certainty level as measured by an entailment-based Agreement Score (AS). Through extensive experiments across 3 KGDG datasets, 3 decoding methods, and 4 KGDG models, we validate the effectiveness of CRR for reducing hallucination in KGDG task.
Authors: Geyang Guo, Ranchi Zhao, Tianyi Tang, Wayne Xin Zhao, Ji-Rong Wen
Abstract: Alignment with human preference is a desired property of large language models (LLMs). Currently, the main alignment approach is based on reinforcement learning from human feedback (RLHF). Despite the effectiveness of RLHF, it is intricate to implement and train, thus recent studies explore how to develop alternative alignment approaches based on supervised fine-tuning (SFT). A major limitation of SFT is that it essentially does imitation learning, which cannot fully understand what are the expected behaviors. To address this issue, we propose an improved alignment approach named FIGA. Different from prior methods, we incorporate fine-grained (i.e., token or phrase level) quality signals that are derived by contrasting good and bad responses. Our approach has made two major contributions. Firstly, we curate a refined alignment dataset that pairs initial responses and the corresponding revised ones. Secondly, we devise a new loss function can leverage fine-grained quality signals to instruct the learning of LLMs for alignment. Extensive experiments have demonstrated the effectiveness of our approaches by comparing a number of competitive baselines.
Authors: Jinyan Su, Claire Cardie, Preslav Nakov
Abstract: In the age of large language models (LLMs) and the widespread adoption of AI-driven content creation, the landscape of information dissemination has witnessed a paradigm shift. With the proliferation of both human-written and machine-generated real and fake news, robustly and effectively discerning the veracity of news articles has become an intricate challenge. While substantial research has been dedicated to fake news detection, this either assumes that all news articles are human-written or abruptly assumes that all machine-generated news are fake. Thus, a significant gap exists in understanding the interplay between machine-(paraphrased) real news, machine-generated fake news, human-written fake news, and human-written real news. In this paper, we study this gap by conducting a comprehensive evaluation of fake news detectors trained in various scenarios. Our primary objectives revolve around the following pivotal question: How to adapt fake news detectors to the era of LLMs? Our experiments reveal an interesting pattern that detectors trained exclusively on human-written articles can indeed perform well at detecting machine-generated fake news, but not vice versa. Moreover, due to the bias of detectors against machine-generated texts \cite{su2023fake}, they should be trained on datasets with a lower machine-generated news ratio than the test set. Building on our findings, we provide a practical strategy for the development of robust fake news detectors.
Authors: Yassir Fathullah, Chunyang Wu, Egor Lakomkin, Ke Li, Junteng Jia, Yuan Shangguan, Jay Mahadeokar, Ozlem Kalinli, Christian Fuegen, Mike Seltzer
Abstract: In this work, we extend the instruction-tuned Llama-2 model with end-to-end general-purpose speech processing and reasoning abilities while maintaining the wide range of original LLM capabilities, without using any carefully curated paired data. The resulting end-to-end model, named AudioChatLlama, can utilize audio prompts as a replacement for text and sustain a conversation. Such a model also has extended cross-modal capabilities such as being able to perform spoken question answering (QA), speech translation, and audio summarization amongst many other closed and open-domain tasks. This is unlike prior approaches in speech, in which LLMs are extended to handle audio for a limited number of pre-designated tasks. On both synthesized and recorded speech QA test sets, evaluations show that our end-to-end approach is on par with or outperforms cascaded systems (speech recognizer + LLM) in terms of modeling the response to a prompt. Furthermore, unlike cascades, our approach can interchange text and audio modalities and intrinsically utilize prior context in a conversation to provide better results.
Authors: Kexin Huang, Xiangyang Liu, Qianyu Guo, Tianxiang Sun, Jiawei Sun, Yaru Wang, Zeyang Zhou, Yixu Wang, Yan Teng, Xipeng Qiu, Yingchun Wang, Dahua Lin
Abstract: The widespread adoption of large language models (LLMs) across various regions underscores the urgent need to evaluate their alignment with human values. Current benchmarks, however, fall short of effectively uncovering safety vulnerabilities in LLMs. Despite numerous models achieving high scores and 'topping the chart' in these evaluations, there is still a significant gap in LLMs' deeper alignment with human values and achieving genuine harmlessness. To this end, this paper proposes a value alignment benchmark named Flames, which encompasses both common harmlessness principles and a unique morality dimension that integrates specific Chinese values such as harmony. Accordingly, we carefully design adversarial prompts that incorporate complex scenarios and jailbreaking methods, mostly with implicit malice. By prompting 17 mainstream LLMs, we obtain model responses and rigorously annotate them for detailed evaluation. Our findings indicate that all the evaluated LLMs demonstrate relatively poor performance on Flames, particularly in the safety and fairness dimensions. We also develop a lightweight specified scorer capable of scoring LLMs across multiple dimensions to efficiently evaluate new models on the benchmark. The complexity of Flames has far exceeded existing benchmarks, setting a new challenge for contemporary LLMs and highlighting the need for further alignment of LLMs. Our benchmark is publicly available at https://github.com/AIFlames/Flames.
Authors: Tatsuki Kuribayashi, Yohei Oseki, Timothy Baldwin
Abstract: Instruction tuning aligns the response of large language models (LLMs) with human preferences. Despite such efforts in human--LLM alignment, we find that instruction tuning does not always make LLMs human-like from a cognitive modeling perspective. More specifically, next-word probabilities estimated by instruction-tuned LLMs are often worse at simulating human reading behavior than those estimated by base LLMs. In addition, we explore prompting methodologies for simulating human reading behavior with LLMs. Our results show that prompts reflecting a particular linguistic hypothesis improve psychometric predictive power, but are still inferior to small base models. These findings highlight that recent advancements in LLMs, i.e., instruction tuning and prompting, do not offer better estimates than direct probability measurements from base LLMs in cognitive modeling. In other words, pure next-word probability remains a strong predictor for human reading behavior, even in the age of LLMs.
Authors: Yahan Yang, Soham Dan, Dan Roth, Insup Lee
Abstract: Multilingual pre-trained Large Language Models (LLMs) are incredibly effective at Question Answering (QA), a core task in Natural Language Understanding, achieving high accuracies on several multilingual benchmarks. However, little is known about how well their confidences are calibrated. In this paper, we comprehensively benchmark the calibration of several multilingual LLMs (MLLMs) on a variety of QA tasks. We perform extensive experiments, spanning encoder-only, encoder-decoder, and decoder-only QA models (size varying from 110M to 7B parameters) and diverse languages, including both high- and low-resource ones. We study different dimensions of calibration in in-distribution, out-of-distribution, and cross-lingual transfer settings, and investigate strategies to improve it, including post-hoc methods and regularized fine-tuning. For decoder-only LLMs such as LlaMa2, we additionally find that in-context learning improves confidence calibration on multilingual data. We also conduct several ablation experiments to study the effect of language distances, language corpus size, and model size on calibration, and how multilingual models compare with their monolingual counterparts for diverse tasks and languages. Our experiments suggest that the multilingual QA models are poorly calibrated for languages other than English and incorporating a small set of cheaply translated multilingual samples during fine-tuning/calibration effectively enhances the calibration performance.
Authors: Minqian Liu, Ying Shen, Zhiyang Xu, Yixin Cao, Eunah Cho, Vaibhav Kumar, Reza Ghanadan, Lifu Huang
Abstract: Natural Language Generation (NLG) typically involves evaluating the generated text in various aspects (e.g., consistency and naturalness) to obtain a comprehensive assessment. However, multi-aspect evaluation remains challenging as it may require the evaluator to generalize to any given evaluation aspect even if it's absent during training. In this paper, we introduce X-Eval, a two-stage instruction tuning framework to evaluate the text in both seen and unseen aspects customized by end users. X-Eval consists of two learning stages: the vanilla instruction tuning stage that improves the model's ability to follow evaluation instructions, and an enhanced instruction tuning stage that exploits the connections between fine-grained evaluation aspects to better assess text quality. To support the training of X-Eval, we collect AspectInstruct, the first instruction tuning dataset tailored for multi-aspect NLG evaluation spanning 27 diverse evaluation aspects with 65 tasks. To enhance task diversity, we devise an augmentation strategy that converts human rating annotations into diverse forms of NLG evaluation tasks, including scoring, comparison, ranking, and Boolean question answering. Extensive experiments across three essential categories of NLG tasks: dialogue generation, summarization, and data-to-text coupled with 21 aspects in meta-evaluation, demonstrate that our X-Eval enables even a lightweight language model to achieve a comparable if not higher correlation with human judgments compared to the state-of-the-art NLG evaluators, such as GPT-4.
Authors: Jierui Li, Vipul Raheja, Dhruv Kumar
Abstract: In recent times, large language models (LLMs) have shown impressive performance on various document-level tasks such as document classification, summarization, and question-answering. However, research on understanding their capabilities on the task of self-contradictions in long documents has been very limited. In this work, we introduce ContraDoc, the first human-annotated dataset to study self-contradictions in long documents across multiple domains, varying document lengths, self-contradictions types, and scope. We then analyze the current capabilities of four state-of-the-art open-source and commercially available LLMs: GPT3.5, GPT4, PaLM2, and LLaMAv2 on this dataset. While GPT4 performs the best and can outperform humans on this task, we find that it is still unreliable and struggles with self-contradictions that require more nuance and context. We release the dataset and all the code associated with the experiments (https://github.com/ddhruvkr/CONTRADOC).
Authors: Lucas Torroba Hennigen, Shannon Shen, Aniruddha Nrusimha, Bernhard Gapp, David Sontag, Yoon Kim
Abstract: LLMs are vulnerable to hallucinations, and thus their outputs generally require laborious human verification for high-stakes applications. To this end, we propose symbolically grounded generation (SymGen) as a simple approach for enabling easier manual validation of an LLM's output. SymGen prompts an LLM to interleave its regular output text with explicit symbolic references to fields present in some conditioning data (e.g., a table in JSON format). The references can be used to display the provenance of different spans of text in the generation, reducing the effort required for manual verification. Across a range of data-to-text and question-answering experiments, we find that LLMs are able to directly output text that makes use of accurate symbolic references while maintaining fluency and factuality. In a human study we further find that such annotations can streamline human verification of machine-generated text. Our code will be available at http://symgen.github.io.
URLs: http://symgen.github.io.
Authors: Adithya Renduchintala, Tugrul Konuk, Oleksii Kuchaiev
Abstract: We introduce Tied-LoRA, a novel paradigm leveraging weight tying and selective training to enhance the parameter efficiency of Low-rank Adaptation (LoRA). Our exploration encompasses different plausible combinations of parameter training and freezing, coupled with weight tying, aimed at identifying the optimal trade-off between performance and the count of trainable parameters. Across $5$ diverse tasks and two foundational language models with different parameter counts, our experiments provide comprehensive insights into the inherent trade-offs between efficiency and performance. Our findings reveal a specific Tied-LoRA configuration that distinguishes itself by showcasing comparable performance to LoRA across multiple tasks while utilizing only a fraction of the parameters employed by the standard LoRA method, particularly at elevated ranks. This underscores the efficacy of Tied-LoRA in achieving impressive results with significantly reduced model complexity.
Authors: Fuxiao Liu, Xiaoyang Wang, Wenlin Yao, Jianshu Chen, Kaiqiang Song, Sangwoo Cho, Yaser Yacoob, Dong Yu
Abstract: With the rapid development of large language models (LLMs) and their integration into large multimodal models (LMMs), there has been impressive progress in zero-shot completion of user-oriented vision-language tasks. However, a gap remains in the domain of chart image understanding due to the distinct abstract components in charts. To address this, we introduce a large-scale MultiModal Chart Instruction (\textbf{MMC-Instruction}) dataset comprising 600k instances supporting diverse tasks and chart types. Leveraging this data, we develop MultiModal Chart Assistant (\textbf{MMCA}), an LMM that achieves state-of-the-art performance on existing chart QA benchmarks. Recognizing the need for a comprehensive evaluation of LMM chart understanding, we also propose a MultiModal Chart Benchmark (\textbf{MMC-Benchmark}), a comprehensive human-annotated benchmark with nine distinct tasks evaluating reasoning capabilities over charts. Extensive experiments on MMC-Benchmark reveal the limitations of existing LMMs on correctly interpreting charts, even for the most recent GPT-4V model. Our work provides an instruction-tuning methodology and benchmark to advance multimodal understanding of charts. Code and data are available at https://github.com/FuxiaoLiu/MMC.
Authors: Aman Kassahun Wassie
Abstract: Machine translation (MT) for low-resource languages such as Ge'ez, an ancient language that is no longer the native language of any community, faces challenges such as out-of-vocabulary words, domain mismatches, and lack of sufficient labeled training data. In this work, we explore various methods to improve Ge'ez MT, including transfer-learning from related languages, optimizing shared vocabulary and token segmentation approaches, finetuning large pre-trained models, and using large language models (LLMs) for few-shot translation with fuzzy matches. We develop a multilingual neural machine translation (MNMT) model based on languages relatedness, which brings an average performance improvement of about 4 BLEU compared to standard bilingual models. We also attempt to finetune the NLLB-200 model, one of the most advanced translation models available today, but find that it performs poorly with only 4k training samples for Ge'ez. Furthermore, we experiment with using GPT-3.5, a state-of-the-art LLM, for few-shot translation with fuzzy matches, which leverages embedding similarity-based retrieval to find context examples from a parallel corpus. We observe that GPT-3.5 achieves a remarkable BLEU score of 9.2 with no initial knowledge of Ge'ez, but still lower than the MNMT baseline of 15.2. Our work provides insights into the potential and limitations of different approaches for low-resource and ancient language MT.
Authors: Zhouhan Lin, Cheng Deng, Le Zhou, Tianhang Zhang, Yi Xu, Yutong Xu, Zhongmou He, Yuanyuan Shi, Beiya Dai, Yunchong Song, Boyi Zeng, Qiyuan Chen, Yuxun Miao, Bo Xue, Shu Wang, Luoyi Fu, Weinan Zhang, Junxian He, Yunqiang Zhu, Xinbing Wang, Chenghu Zhou
Abstract: Large language models (LLMs) have achieved huge success for their general knowledge and ability to solve a wide spectrum of tasks in natural language processing (NLP). Due to their impressive abilities, LLMs have shed light on potential inter-discipline applications to foster scientific discoveries of a specific domain by using artificial intelligence (AI for science, AI4S). In the meantime, utilizing NLP techniques in geoscience research and practice is wide and convoluted, contributing from knowledge extraction and document classification to question answering and knowledge discovery. In this work, we take the initial step to leverage LLM for science, through a rather straightforward approach. We try to specialize an LLM into geoscience, by further pre-training the model with a vast amount of texts in geoscience, as well as supervised fine-tuning (SFT) the resulting model with our custom collected instruction tuning dataset. These efforts result in a model GeoGalactica consisting of 30 billion parameters. To our best knowledge, it is the largest language model for the geoscience domain. More specifically, GeoGalactica is from further pre-training of Galactica. We train GeoGalactica over a geoscience-related text corpus containing 65 billion tokens, preserving as the largest geoscience-specific text corpus. Then we fine-tune the model with 1 million pairs of instruction-tuning data consisting of questions that demand professional geoscience knowledge to answer. In this technical report, we will illustrate in detail all aspects of GeoGalactica, including data collection, data cleaning, base model selection, pre-training, SFT, and evaluation. We open-source our data curation tools and the checkpoints of GeoGalactica during the first 3/4 of pre-training.
Authors: Shuaijie She, Wei Zou, Shujian Huang, Wenhao Zhu, Xiang Liu, Xiang Geng, Jiajun Chen
Abstract: Though reasoning abilities are considered language-agnostic, existing LLMs exhibit inconsistent reasoning abilities across different languages, e.g., reasoning in the dominant language like English is superior to other languages due to the imbalance of multilingual training data. To enhance reasoning abilities in non-dominant languages, we propose a Multilingual-Alignment-as-Preference Optimization framework (MAPO), aiming to align the reasoning processes in other languages with the dominant language. Specifically, we harness an off-the-shelf translation model for the consistency between answers in non-dominant and dominant languages, which we adopt as the preference for optimization, e.g., Direct Preference Optimization (DPO) or Proximal Policy Optimization (PPO). Experiments show that MAPO stably achieves significant improvements in the multilingual reasoning of various models on all three benchmarks (MSVAMP +16.2%, MGSM +6.1%, and MNumGLUESub +13.3%), with improved reasoning consistency across languages.
Authors: Alisa Liu, Xiaochuang Han, Yizhong Wang, Yulia Tsvetkov, Yejin Choi, Noah A. Smith
Abstract: Despite the general capabilities of large pretrained language models, they consistently benefit from further adaptation to better achieve desired behaviors. However, tuning these models has become increasingly resource-intensive, or impossible when model weights are private. We introduce proxy-tuning, a lightweight decoding-time algorithm that operates on top of black-box LMs to achieve the same end as direct tuning, but by accessing only its predictions over the output vocabulary, not its parameters. Our method tunes a smaller LM, then applies the difference between the predictions of the small tuned and untuned LMs to shift the original predictions of the larger untuned model in the direction of tuning, while retaining the benefits of larger-scale pretraining. In experiments, when we apply proxy-tuning to Llama2-70B using proxies of only 7B size, we can close 88% of the gap between Llama2-70B and its truly-tuned chat version, when evaluated across knowledge, reasoning, and safety benchmarks. Interestingly, on TruthfulQA, proxy-tuned models are actually more truthful than directly tuned models, possibly because decoding-time guidance better retains the model's factual knowledge. We then demonstrate the generality of proxy-tuning by applying it to domain adaptation on code, and task-specific finetuning on question-answering and math problems. Finally, we show how to proxy-tune a truly black-box LM, GPT-3.5, for temporal adaptation, increasing its knowledge about recent events. Our work demonstrates the promise of using small tuned LMs to efficiently customize large, potentially proprietary LMs through decoding-time guidance.
Authors: Chengzhi Mao, Carl Vondrick, Hao Wang, Junfeng Yang
Abstract: We find that large language models (LLMs) are more likely to modify human-written text than AI-generated text when tasked with rewriting. This tendency arises because LLMs often perceive AI-generated text as high-quality, leading to fewer modifications. We introduce a method to detect AI-generated content by prompting LLMs to rewrite text and calculating the editing distance of the output. We dubbed our geneRative AI Detection viA Rewriting method Raidar. Raidar significantly improves the F1 detection scores of existing AI content detection models -- both academic and commercial -- across various domains, including News, creative writing, student essays, code, Yelp reviews, and arXiv papers, with gains of up to 29 points. Operating solely on word symbols without high-dimensional features, our method is compatible with black box LLMs, and is inherently robust on new content. Our results illustrate the unique imprint of machine-generated text through the lens of the machines themselves.
Authors: Lang Cao, Zifeng Wang, Cao Xiao, Jimeng Sun
Abstract: Machine learning shows promise in predicting the outcome of legal cases, but most research has concentrated on civil law cases rather than case law systems. We identified two unique challenges in making legal case outcome predictions with case law. First, it is crucial to identify relevant precedent cases that serve as fundamental evidence for judges during decision-making. Second, it is necessary to consider the evolution of legal principles over time, as early cases may adhere to different legal contexts. In this paper, we proposed a new framework named PILOT (PredictIng Legal case OuTcome) for case outcome prediction. It comprises two modules for relevant case retrieval and temporal pattern handling, respectively. To benchmark the performance of existing legal case outcome prediction models, we curated a dataset from a large-scale case law database. We demonstrate the importance of accurately identifying precedent cases and mitigating the temporal shift when making predictions for case law, as our method shows a significant improvement over the prior methods that focus on civil law case outcome predictions.
Authors: Philipp Singer, Pascal Pfeiffer, Yauhen Babakhin, Maximilian Jeblick, Nischay Dhankhar, Gabor Fodor, Sri Satish Ambati
Abstract: We present H2O-Danube, a series of small 1.8B language models consisting of H2O-Danube-1.8B, trained on 1T tokens, and the incremental improved H2O-Danube2-1.8B trained on an additional 2T tokens. Our models exhibit highly competitive metrics across a multitude of benchmarks and, as of the time of this writing, H2O-Danube2-1.8B achieves the top ranking on Open LLM Leaderboard for all models below the 2B parameter range. The models follow core principles of LLama 2 and Mistral, and we leverage and refine various techniques for pre-training large language models. We additionally release chat models trained with supervised fine-tuning followed by direct preference optimization. We make all models openly available under Apache 2.0 license further democratizing LLMs to a wider audience economically.
Authors: Xue-Yong Fu, Md Tahmid Rahman Laskar, Elena Khasanova, Cheng Chen, Shashi Bhushan TN
Abstract: Large Language Models (LLMs) have demonstrated impressive capabilities to solve a wide range of tasks without being explicitly fine-tuned on task-specific datasets. However, deploying LLMs in the real world is not trivial, as it requires substantial computing resources. In this paper, we investigate whether smaller, compact LLMs are a good alternative to the comparatively Larger LLMs2 to address significant costs associated with utilizing LLMs in the real world. In this regard, we study the meeting summarization task in a real-world industrial environment and conduct extensive experiments by comparing the performance of fine-tuned compact LLMs (e.g., FLAN-T5, TinyLLaMA, LiteLLaMA) with zero-shot larger LLMs (e.g., LLaMA-2, GPT-3.5, PaLM-2). We observe that most smaller LLMs, even after fine-tuning, fail to outperform larger zero-shot LLMs in meeting summarization datasets. However, a notable exception is FLAN-T5 (780M parameters), which performs on par or even better than many zero-shot Larger LLMs (from 7B to above 70B parameters), while being significantly smaller. This makes compact LLMs like FLAN-T5 a suitable cost-efficient solution for real-world industrial deployment.
Authors: Md. Kowsher, Md. Shohanur Islam Sobuj, Asif Mahmud, Nusrat Jahan Prottasha, Prakash Bhat
Abstract: Efficiently fine-tuning Large Language Models (LLMs) for specific tasks presents a considerable challenge in natural language processing. Traditional methods, like prompt or prefix tuning, typically rely on arbitrary tokens for training, leading to prolonged training times and generalized token use across various class labels. To address these issues, this paper introduces L-Tuning, an efficient fine-tuning approach designed for classification tasks within the Natural Language Inference (NLI) framework. Diverging from conventional methods, L-Tuning focuses on the fine-tuning of label tokens processed through a pre-trained LLM, thereby harnessing its pre-existing semantic knowledge. This technique not only improves the fine-tuning accuracy and efficiency but also facilitates the generation of distinct label embeddings for each class, enhancing the model's training nuance. Our experimental results indicate a significant improvement in training efficiency and classification accuracy with L-Tuning compared to traditional approaches, marking a promising advancement in fine-tuning LLMs for complex language tasks.
Authors: Chang Zong, Yuchen Yan, Weiming Lu, Eliot Huang, Jian Shao, Yueting Zhuang
Abstract: Recent progress with LLM-based agents has shown promising results across various tasks. However, their use in answering questions from knowledge bases remains largely unexplored. Implementing a KBQA system using traditional methods is challenging due to the shortage of task-specific training data and the complexity of creating task-focused model structures. In this paper, we present Triad, a unified framework that utilizes an LLM-based agent with three roles for KBQA tasks. The agent is assigned three roles to tackle different KBQA subtasks: agent as a generalist for mastering various subtasks, as a decision maker for the selection of candidates, and as an advisor for answering questions with knowledge. Our KBQA framework is executed in four phases, involving the collaboration of the agent's multiple roles. We evaluated the performance of our framework using three benchmark datasets, and the results show that our framework outperforms state-of-the-art systems on the LC-QuAD and YAGO-QA benchmarks, yielding F1 scores of 11.8% and 20.7%, respectively.
Authors: Nikhil Narayan, Mrutyunjay Biswal
Abstract: In the digital realm, rich data serves as a crucial source of insights into the complexities of social, political, and economic landscapes. Addressing the growing need for high-quality information on events and the imperative to combat hate speech, this research led to the establishment of the Shared Task on Climate Activism Stance and Hate Event Detection at CASE 2024. Focused on climate activists contending with hate speech on social media, our study contributes to hate speech identification from tweets. Analyzing three sub-tasks - Hate Speech Detection (Sub-task A), Targets of Hate Speech Identification (Sub-task B), and Stance Detection (Sub-task C) - Team Z-AGI Labs evaluated various models, including LSTM, Xgboost, and LGBM based on Tf-Idf. Results unveiled intriguing variations, with Catboost excelling in Subtask-B (F1: 0.5604) and Subtask-C (F1: 0.7081), while LGBM emerged as the top-performing model for Subtask-A (F1: 0.8684). This research provides valuable insights into the suitability of classical machine learning models for climate hate speech and stance detection, aiding informed model selection for robust mechanisms.
Authors: Huixuan Zhang, Junzhe Zhang, Xiaojun Wan
Abstract: Large-scale vision-language models have demonstrated impressive skill in handling tasks that involve both areas. Nevertheless, these models frequently experience significant issues with generating inaccurate information, which is hallucination. In this study, we concentrate on a specific type of hallucination-number hallucination, referring to models incorrectly identifying the number of certain objects in pictures. We perform quantitative evaluations regarding number hallucination, showing it to be critical in major open-source large vision-language models. Furthermore, we utilizes two related tasks to conduct an in-depth analysis of number hallucination, revealing the severe inner and outer inconsistency among all tasks. Based on this examination, we devise a training approach aimed at improving consistency to reduce number hallucinations, which leads to an 8% enhancement in performance over direct finetuning methods. Our code and dataset will be released to the community.
Authors: Siqi Fan, Xin Jiang, Xiang Li, Xuying Meng, Peng Han, Shuo Shang, Aixin Sun, Yequan Wang, Zhongyuan Wang
Abstract: The inference phase of Large Language Models (LLMs) is very expensive. An ideal inference stage of LLMs could utilize fewer computational resources while still maintaining its capabilities (e.g., generalization and in-context learning ability). In this paper, we try to answer the question, "During LLM inference, can we use shallow layers for easy instances; and deep layers for hard ones?" To answer this question, we first indicate that Not all Layers are Necessary during Inference by statistically analyzing the activated layers across tasks. Then, we propose a simple algorithm named AdaInfer to determine the inference termination moment based on the input instance adaptively. More importantly, AdaInfer does not alter LLM parameters and maintains generalizability across tasks. Experiments on well-known LLMs (i.e., Llama2 series and OPT) show that AdaInfer saves an average of 14.8% of computational resources, even up to 50% on sentiment tasks, while maintaining comparable performance. Additionally, this method is orthogonal to other model acceleration techniques, potentially boosting inference efficiency further.
Authors: Shangjian Yin, Peijie Huang, Yuhong Xu, Haojing Huang, Jiatian Chen
Abstract: 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}.
Authors: Qiao Sun, Liujia Yang, Minghao Ma, Nanyang Ye, Qinying Gu
Abstract: Aspect Sentiment Triplet Extraction (ASTE) is a burgeoning subtask of fine-grained sentiment analysis, aiming to extract structured sentiment triplets from unstructured textual data. Existing approaches to ASTE often complicate the task with additional structures or external data. In this research, we propose a novel tagging scheme and employ a contrastive learning approach to mitigate these challenges. The proposed approach demonstrates comparable or superior performance in comparison to state-of-the-art techniques, while featuring a more compact design and reduced computational overhead. Notably, even in the era of Large Language Models (LLMs), our method exhibits superior efficacy compared to GPT 3.5 and GPT 4 in a few-shot learning scenarios. This study also provides valuable insights for the advancement of ASTE techniques within the paradigm of large language models.
Authors: Joonwon Jang, Sanghwan Jang, Wonbin Kweon, Minjin Jeon, Hwanjo Yu
Abstract: Large language models (LLMs) are able to solve various tasks with only a few demonstrations utilizing their in-context learning (ICL) abilities. However, LLMs often rely on their pre-trained semantic priors of demonstrations rather than on the input-label relationships to proceed with ICL prediction. In this work, we term this phenomenon as the 'Demonstration Shortcut'. While previous works have primarily focused on improving ICL prediction results for predefined tasks, we aim to rectify the Demonstration Shortcut, thereby enabling the LLM to effectively learn new input-label relationships from demonstrations. To achieve this, we introduce In-Context Calibration, a demonstration-aware calibration method. We evaluate the effectiveness of the proposed method in two settings: (1) the Original ICL Task using the standard label space and (2) the Task Learning setting, where the label space is replaced with semantically unrelated tokens. In both settings, In-Context Calibration demonstrates substantial improvements, with results generalized across three LLM families (OPT, GPT, and Llama2) under various configurations.
Authors: Jiaxin Guo, Hao Yang, Zongyao Li, Daimeng Wei, Hengchao Shang, Xiaoyu Chen
Abstract: This paper presents a study on strategies to enhance the translation capabilities of large language models (LLMs) in the context of machine translation (MT) tasks. The paper proposes a novel paradigm consisting of three stages: Secondary Pre-training using Extensive Monolingual Data, Continual Pre-training with Interlinear Text Format Documents, and Leveraging Source-Language Consistent Instruction for Supervised Fine-Tuning. Previous research on LLMs focused on various strategies for supervised fine-tuning (SFT), but their effectiveness has been limited. While traditional machine translation approaches rely on vast amounts of parallel bilingual data, our paradigm highlights the importance of using smaller sets of high-quality bilingual data. We argue that the focus should be on augmenting LLMs' cross-lingual alignment abilities during pre-training rather than solely relying on extensive bilingual data during SFT. Experimental results conducted using the Llama2 model, particularly on Chinese-Llama2 after monolingual augmentation, demonstrate the improved translation capabilities of LLMs. A significant contribution of our approach lies in Stage2: Continual Pre-training with Interlinear Text Format Documents, which requires less than 1B training data, making our method highly efficient. Additionally, in Stage3, we observed that setting instructions consistent with the source language benefits the supervised fine-tuning process. Experimental results demonstrate that our approach surpasses previous work and achieves superior performance compared to models such as NLLB-54B and GPT3.5-text-davinci-003, despite having a significantly smaller parameter count of only 7B or 13B. This achievement establishes our method as a pioneering strategy in the field of machine translation.
Authors: Zeyu Liu, Souvik Kundu, Anni Li, Junrui Wan, Lianghao Jiang, Peter Anthony Beerel
Abstract: We present a novel Parameter-Efficient Fine-Tuning (PEFT) method, dubbed as Adaptive Freezing of Low Rank Adaptation (AFLoRA). Specifically, for each pre-trained frozen weight tensor, we add a parallel path of trainable low-rank matrices, namely a down-projection and an up-projection matrix, each of which is followed by a feature transformation vector. Based on a novel freezing score, we the incrementally freeze these projection matrices during fine-tuning to reduce the computation and alleviate over-fitting. Our experimental results demonstrate that we can achieve state-of-the-art performance with an average improvement of up to $0.85\%$ as evaluated on GLUE benchmark while yeilding up to $9.5\times$ fewer average trainable parameters. While compared in terms of runtime, AFLoRA can yield up to $1.86\times$ improvement as opposed to similar PEFT alternatives. Besides the practical utility of our approach, we provide insights on the trainability requirements of LoRA paths at different modules and the freezing schedule for the different projection matrices. Code will be released.
Authors: Mengru Wang, Ningyu Zhang, Ziwen Xu, Zekun Xi, Shumin Deng, Yunzhi Yao, Qishen Zhang, Linyi Yang, Jindong Wang, Huajun Chen
Abstract: This paper investigates using knowledge editing techniques to detoxify Large Language Models (LLMs). We construct a benchmark, SafeEdit, which covers nine unsafe categories with various powerful attack prompts and equips comprehensive metrics for systematic evaluation. We conduct experiments with several knowledge editing approaches, indicating that knowledge editing has the potential to efficiently detoxify LLMs with limited impact on general performance. Then, we propose a simple yet effective baseline, dubbed Detoxifying with Intraoperative Neural Monitoring (DINM), to diminish the toxicity of LLMs within a few tuning steps via only one instance. We further provide an in-depth analysis of the internal mechanism for various detoxifying approaches, demonstrating that previous methods like SFT and DPO may merely suppress the activations of toxic parameters, while DINM mitigates the toxicity of the toxic parameters to a certain extent, making permanent adjustments. We hope that these insights could shed light on future work of developing detoxifying approaches and the underlying knowledge mechanisms of LLMs. Code and benchmark are available at https://github.com/zjunlp/EasyEdit.
Authors: Zequan Liu, Jiawen Lyn, Wei Zhu, Xing Tian, Yvette Graham
Abstract: Parameter-efficient fine-tuning (PEFT) is widely studied for its effectiveness and efficiency in the era of large language models. Low-rank adaptation (LoRA) has demonstrated commendable performance as a popular and representative method. However, it is implemented with a fixed intrinsic rank that might not be the ideal setting for the downstream tasks. Recognizing the need for more flexible downstream task adaptation, we extend the methodology of LoRA to an innovative approach we call allocating low-rank adaptation (ALoRA) that enables dynamic adjustments to the intrinsic rank during the adaptation process. First, we propose a novel method, AB-LoRA, that can effectively estimate the importance score of each LoRA rank. Second, guided by AB-LoRA, we gradually prune abundant and negatively impacting LoRA ranks and allocate the pruned LoRA budgets to important Transformer modules needing higher ranks. We have conducted experiments on various tasks, and the experimental results demonstrate that our ALoRA method can outperform the recent baselines with comparable tunable parameters.
Authors: Aleksandra Edwards, Jose Camacho-Collados
Abstract: Recent foundational language models have shown state-of-the-art performance in many NLP tasks in zero- and few-shot settings. An advantage of these models over more standard approaches based on fine-tuning is the ability to understand instructions written in natural language (prompts), which helps them generalise better to different tasks and domains without the need for specific training data. This makes them suitable for addressing text classification problems for domains with limited amounts of annotated instances. However, existing research is limited in scale and lacks understanding of how text generation models combined with prompting techniques compare to more established methods for text classification such as fine-tuning masked language models. In this paper, we address this research gap by performing a large-scale evaluation study for 16 text classification datasets covering binary, multiclass, and multilabel problems. In particular, we compare zero- and few-shot approaches of large language models to fine-tuning smaller language models. We also analyse the results by prompt, classification type, domain, and number of labels. In general, the results show how fine-tuning smaller and more efficient language models can still outperform few-shot approaches of larger language models, which have room for improvement when it comes to text classification.
Authors: Yu Li, Shenyu Zhang, Rui Wu, Xiutian Huang, Yongrui Chen, Wenhao Xu, Guilin Qi, Dehai Min
Abstract: Recent advancements in generative Large Language Models(LLMs) have been remarkable, however, the quality of the text generated by these models often reveals persistent issues. Evaluating the quality of text generated by these models, especially in open-ended text, has consistently presented a significant challenge. Addressing this, recent work has explored the possibility of using LLMs as evaluators. While using a single LLM as an evaluation agent shows potential, it is filled with significant uncertainty and instability. To address these issues, we propose the MATEval: A "Multi-Agent Text Evaluation framework" where all agents are played by LLMs like GPT-4. The MATEval framework emulates human collaborative discussion methods, integrating multiple agents' interactions to evaluate open-ended text. Our framework incorporates self-reflection and Chain-of-Thought (CoT) strategies, along with feedback mechanisms, enhancing the depth and breadth of the evaluation process and guiding discussions towards consensus, while the framework generates comprehensive evaluation reports, including error localization, error types and scoring. Experimental results show that our framework outperforms existing open-ended text evaluation methods and achieves the highest correlation with human evaluation, which confirms the effectiveness and advancement of our framework in addressing the uncertainties and instabilities in evaluating LLMs-generated text. Furthermore, our framework significantly improves the efficiency of text evaluation and model iteration in industrial scenarios.
Authors: Jingzhe Shi, Jialuo Li, Qinwei Ma, Zaiwen Yang, Huan Ma, Lei Li
Abstract: Businesses and software platforms are increasingly turning to Large Language Models (LLMs) such as GPT-3.5, GPT-4, GLM-3, and LLaMa-2 for chat assistance with file access or as reasoning agents for customer service. However, current LLM-based customer service models have limited integration with customer profiles and lack the operational capabilities necessary for effective service. Moreover, existing API integrations emphasize diversity over the precision and error avoidance essential in real-world customer service scenarios. To address these issues, we propose an LLM agent named CHOPS (CHat with custOmer Profile in existing System), designed to: (1) efficiently utilize existing databases or systems for accessing user information or interacting with these systems following existing guidelines; (2) provide accurate and reasonable responses or carry out required operations in the system while avoiding harmful operations; and (3) leverage a combination of small and large LLMs to achieve satisfying performance at a reasonable inference cost. We introduce a practical dataset, the CPHOS-dataset, which includes a database, guiding files, and QA pairs collected from CPHOS, an online platform that facilitates the organization of simulated Physics Olympiads for high school teachers and students. We have conducted extensive experiments to validate the performance of our proposed CHOPS architecture using the CPHOS-dataset, with the aim of demonstrating how LLMs can enhance or serve as alternatives to human customer service. Code for our proposed architecture and dataset can be found at {https://github.com/JingzheShi/CHOPS}.
Authors: Wei Chen, Zhiyuan Li
Abstract: Language models have shown effectiveness in a variety of software applications, particularly in tasks related to automatic workflow. These models possess the crucial ability to call functions, which is essential in creating AI agents. Despite the high performance of large-scale language models in cloud environments, they are often associated with concerns over privacy and cost. Current on-device models for function calling face issues with latency and accuracy. Our research presents a new method that empowers an on-device model with 2 billion parameters to surpass the performance of GPT-4 in both accuracy and latency, and decrease the context length by 95\%. When compared to Llama-7B with a RAG-based function calling mechanism, our method enhances latency by 35-fold. This method reduces the latency to levels deemed suitable for deployment across a variety of edge devices in production environments, aligning with the performance requisites for real-world applications.
Authors: Kang Min Yoo, Jaegeun Han, Sookyo In, Heewon Jeon, Jisu Jeong, Jaewook Kang, Hyunwook Kim, Kyung-Min Kim, Munhyong Kim, Sungju Kim, Donghyun Kwak, Hanock Kwak, Se Jung Kwon, Bado Lee, Dongsoo Lee, Gichang Lee, Jooho Lee, Baeseong Park, Seongjin Shin, Joonsang Yu, Seolki Baek, Sumin Byeon, Eungsup Cho, Dooseok Choe, Jeesung Han, Youngkyun Jin, Hyein Jun, Jaeseung Jung, Chanwoong Kim, Jinhong Kim, Jinuk Kim, Dokyeong Lee, Dongwook Park, Jeong Min Sohn, Sujung Han, Jiae Heo, Sungju Hong, Mina Jeon, Hyunhoon Jung, Jungeun Jung, Wangkyo Jung, Chungjoon Kim, Hyeri Kim, Jonghyun Kim, Min Young Kim, Soeun Lee, Joonhee Park, Jieun Shin, Sojin Yang, Jungsoon Yoon, Hwaran Lee, Sanghwan Bae, Jeehwan Cha, Karl Gylleus, Donghoon Ham, Mihak Hong, Youngki Hong, Yunki Hong, Dahyun Jang, Hyojun Jeon, Yujin Jeon, Yeji Jeong, Myunggeun Ji, Yeguk Jin, Chansong Jo, Shinyoung Joo, Seunghwan Jung, Adrian Jungmyung Kim, Byoung Hoon Kim, Hyomin Kim, Jungwhan Kim, Minkyoung Kim, Minseung Kim, Sungdong Kim, Yonghee Kim, Youngjun Kim, Youngkwan Kim, Donghyeon Ko, Dughyun Lee, Ha Young Lee, Jaehong Lee, Jieun Lee, Jonghyun Lee, Jongjin Lee, Min Young Lee, Yehbin Lee, Taehong Min, Yuri Min, Kiyoon Moon, Hyangnam Oh, Jaesun Park, Kyuyon Park, Younghun Park, Hanbae Seo, Seunghyun Seo, Mihyun Sim, Gyubin Son, Matt Yeo, Kyung Hoon Yeom, Wonjoon Yoo, Myungin You, Doheon Ahn, Homin Ahn, Joohee Ahn, Seongmin Ahn, Chanwoo An, Hyeryun An, Junho An, Sang-Min An, Boram Byun, Eunbin Byun, Jongho Cha, Minji Chang, Seunggyu Chang, Haesong Cho, Youngdo Cho, Dalnim Choi, Daseul Choi, Hyoseok Choi, Minseong Choi, Sangho Choi, Seongjae Choi, Wooyong Choi, Sewhan Chun, Dong Young Go, Chiheon Ham, Danbi Han, Jaemin Han, Moonyoung Hong, Sung Bum Hong, Dong-Hyun Hwang, Seongchan Hwang, Jinbae Im, Hyuk Jin Jang, Jaehyung Jang, Jaeni Jang, Sihyeon Jang, Sungwon Jang, Joonha Jeon, Daun Jeong, Joonhyun Jeong, Kyeongseok Jeong, Mini Jeong, Sol Jin, Hanbyeol Jo, Hanju Jo, Minjung Jo, Chaeyoon Jung, Hyungsik Jung, Jaeuk Jung, Ju Hwan Jung, Kwangsun Jung, Seungjae Jung, Soonwon Ka, Donghan Kang, Soyoung Kang, Taeho Kil, Areum Kim, Beomyoung Kim, Byeongwook Kim, Daehee Kim, Dong-Gyun Kim, Donggook Kim, Donghyun Kim, Euna Kim, Eunchul Kim, Geewook Kim, Gyu Ri Kim, Hanbyul Kim, Heesu Kim, Isaac Kim, Jeonghoon Kim, Jihye Kim, Joonghoon Kim, Minjae Kim, Minsub Kim, Pil Hwan Kim, Sammy Kim, Seokhun Kim, Seonghyeon Kim, Soojin Kim, Soong Kim, Soyoon Kim, Sunyoung Kim, Taeho Kim, Wonho Kim, Yoonsik Kim, You Jin Kim, Yuri Kim, Beomseok Kwon, Ohsung Kwon, Yoo-Hwan Kwon, Anna Lee, Byungwook Lee, Changho Lee, Daun Lee, Dongjae Lee, Ha-Ram Lee, Hodong Lee, Hwiyeong Lee, Hyunmi Lee, Injae Lee, Jaeung Lee, Jeongsang Lee, Jisoo Lee, Jongsoo Lee, Joongjae Lee, Juhan Lee, Jung Hyun Lee, Junghoon Lee, Junwoo Lee, Se Yun Lee, Sujin Lee, Sungjae Lee, Sungwoo Lee, Wonjae Lee, Zoo Hyun Lee, Jong Kun Lim, Kun Lim, Taemin Lim, Nuri Na, Jeongyeon Nam, Kyeong-Min Nam, Yeonseog Noh, Biro Oh, Jung-Sik Oh, Solgil Oh, Yeontaek Oh, Boyoun Park, Cheonbok Park, Dongju Park, Hyeonjin Park, Hyun Tae Park, Hyunjung Park, Jihye Park, Jooseok Park, Junghwan Park, Jungsoo Park, Miru Park, Sang Hee Park, Seunghyun Park, Soyoung Park, Taerim Park, Wonkyeong Park, Hyunjoon Ryu, Jeonghun Ryu, Nahyeon Ryu, Soonshin Seo, Suk Min Seo, Yoonjeong Shim, Kyuyong Shin, Wonkwang Shin, Hyun Sim, Woongseob Sim, Hyejin Soh, Bokyong Son, Hyunjun Son, Seulah Son, Chi-Yun Song, Chiyoung Song, Ka Yeon Song, Minchul Song, Seungmin Song, Jisung Wang, Yonggoo Yeo, Myeong Yeon Yi, Moon Bin Yim, Taehwan Yoo, Youngjoon Yoo, Sungmin Yoon, Young Jin Yoon, Hangyeol Yu, Ui Seon Yu, Xingdong Zuo, Jeongin Bae, Joungeun Bae, Hyunsoo Cho, Seonghyun Cho, Yongjin Cho, Taekyoon Choi, Yera Choi, Jiwan Chung, Zhenghui Han, Byeongho Heo, Euisuk Hong, Taebaek Hwang, Seonyeol Im, Sumin Jegal, Sumin Jeon, Yelim Jeong, Yonghyun Jeong, Can Jiang, Juyong Jiang, Jiho Jin, Ara Jo, Younghyun Jo, Hoyoun Jung, Juyoung Jung, Seunghyeong Kang, Dae Hee Kim, Ginam Kim, Hangyeol Kim, Heeseung Kim, Hyojin Kim, Hyojun Kim, Hyun-Ah Kim, Jeehye Kim, Jin-Hwa Kim, Jiseon Kim, Jonghak Kim, Jung Yoon Kim, Rak Yeong Kim, Seongjin Kim, Seoyoon Kim, Sewon Kim, Sooyoung Kim, Sukyoung Kim, Taeyong Kim, Naeun Ko, Bonseung Koo, Heeyoung Kwak, Haena Kwon, Youngjin Kwon, Boram Lee, Bruce W. Lee, Dagyeong Lee, Erin Lee, Euijin Lee, Ha Gyeong Lee, Hyojin Lee, Hyunjeong Lee, Jeeyoon Lee, Jeonghyun Lee, Jongheok Lee, Joonhyung Lee, Junhyuk Lee, Mingu Lee, Nayeon Lee, Sangkyu Lee, Se Young Lee, Seulgi Lee, Seung Jin Lee, Suhyeon Lee, Yeonjae Lee, Yesol Lee, Youngbeom Lee, Yujin Lee, Shaodong Li, Tianyu Liu, Seong-Eun Moon, Taehong Moon, Max-Lasse Nihlenramstroem, Wonseok Oh, Yuri Oh, Hongbeen Park, Hyekyung Park, Jaeho Park, Nohil Park, Sangjin Park, Jiwon Ryu, Miru Ryu, Simo Ryu, Ahreum Seo, Hee Seo, Kangdeok Seo, Jamin Shin, Seungyoun Shin, Heetae Sin, Jiangping Wang, Lei Wang, Ning Xiang, Longxiang Xiao, Jing Xu, Seonyeong Yi, Haanju Yoo, Haneul Yoo, Hwanhee Yoo, Liang Yu, Youngjae Yu, Weijie Yuan, Bo Zeng, Qian Zhou, Kyunghyun Cho, Jung-Woo Ha, Joonsuk Park, Jihyun Hwang, Hyoung Jo Kwon, Soonyong Kwon, Jungyeon Lee, Seungho Lee, Seonghyeon Lim, Hyunkyung Noh, Seungho Choi, Sang-Woo Lee, Jung Hwa Lim, Nako Sung
Abstract: We introduce HyperCLOVA X, a family of large language models (LLMs) tailored to the Korean language and culture, along with competitive capabilities in English, math, and coding. HyperCLOVA X was trained on a balanced mix of Korean, English, and code data, followed by instruction-tuning with high-quality human-annotated datasets while abiding by strict safety guidelines reflecting our commitment to responsible AI. The model is evaluated across various benchmarks, including comprehensive reasoning, knowledge, commonsense, factuality, coding, math, chatting, instruction-following, and harmlessness, in both Korean and English. HyperCLOVA X exhibits strong reasoning capabilities in Korean backed by a deep understanding of the language and cultural nuances. Further analysis of the inherent bilingual nature and its extension to multilingualism highlights the model's cross-lingual proficiency and strong generalization ability to untargeted languages, including machine translation between several language pairs and cross-lingual inference tasks. We believe that HyperCLOVA X can provide helpful guidance for regions or countries in developing their sovereign LLMs.
Authors: Zhewei Sun, Qian Hu, Rahul Gupta, Richard Zemel, Yang Xu
Abstract: Recent advancement in large language models (LLMs) has offered a strong potential for natural language systems to process informal language. A representative form of informal language is slang, used commonly in daily conversations and online social media. To date, slang has not been comprehensively evaluated in LLMs due partly to the absence of a carefully designed and publicly accessible benchmark. Using movie subtitles, we construct a dataset that supports evaluation on a diverse set of tasks pertaining to automatic processing of slang. For both evaluation and finetuning, we show the effectiveness of our dataset on two core applications: 1) slang detection, and 2) identification of regional and historical sources of slang from natural sentences. We also show how our dataset can be used to probe the output distributions of LLMs for interpretive insights. We find that while LLMs such as GPT-4 achieve good performance in a zero-shot setting, smaller BERT-like models finetuned on our dataset achieve comparable performance. Furthermore, we show that our dataset enables finetuning of LLMs such as GPT-3.5 that achieve substantially better performance than strong zero-shot baselines. Our work offers a comprehensive evaluation and a high-quality benchmark on English slang based on the OpenSubtitles corpus, serving both as a publicly accessible resource and a platform for applying tools for informal language processing.
Authors: Yuri Bizzoni, Pascale Feldkamp, Ida Marie Lassen, Mia Jacobsen, Mads Rosendahl Thomsen, Kristoffer Nielbo
Abstract: In this study, we employ a classification approach to show that different categories of literary "quality" display unique linguistic profiles, leveraging a corpus that encompasses titles from the Norton Anthology, Penguin Classics series, and the Open Syllabus project, contrasted against contemporary bestsellers, Nobel prize winners and recipients of prestigious literary awards. Our analysis reveals that canonical and so called high-brow texts exhibit distinct textual features when compared to other quality categories such as bestsellers and popular titles as well as to control groups, likely responding to distinct (but not mutually exclusive) models of quality. We apply a classic machine learning approach, namely Random Forest, to distinguish quality novels from "control groups", achieving up to 77\% F1 scores in differentiating between the categories. We find that quality category tend to be easier to distinguish from control groups than from other quality categories, suggesting than literary quality features might be distinguishable but shared through quality proxies.
Authors: Pardis Moradbeiki, Nasser Ghadiri
Abstract: Smart cities need the involvement of their residents to enhance quality of life. Conversational query-answering is an emerging approach for user engagement. There is an increasing demand of an advanced conversational question-answering that goes beyond classic systems. Existing approaches have shown that LLMs offer promising capabilities for CQA, but may struggle to capture the nuances of conversational contexts. The new approach involves understanding the content and engaging in a multi-step conversation with the user to fulfill their needs. This paper presents a novel method to elevate the performance of Persian Conversational question-answering (CQA) systems. It combines the strengths of Large Language Models (LLMs) with contextual keyword extraction. Our method extracts keywords specific to the conversational flow, providing the LLM with additional context to understand the user's intent and generate more relevant and coherent responses. We evaluated the effectiveness of this combined approach through various metrics, demonstrating significant improvements in CQA performance compared to an LLM-only baseline. The proposed method effectively handles implicit questions, delivers contextually relevant answers, and tackles complex questions that rely heavily on conversational context. The findings indicate that our method outperformed the evaluation benchmarks up to 8% higher than existing methods and the LLM-only baseline.
Authors: Yiming Li, Xueqing Peng, Jianfu Li, Xu Zuo, Suyuan Peng, Donghong Pei, Cui Tao, Hua Xu, Na Hong
Abstract: In acupuncture therapy, the accurate location of acupoints is essential for its effectiveness. The advanced language understanding capabilities of large language models (LLMs) like Generative Pre-trained Transformers (GPT) present a significant opportunity for extracting relations related to acupoint locations from textual knowledge sources. This study aims to compare the performance of GPT with traditional deep learning models (Long Short-Term Memory (LSTM) and Bidirectional Encoder Representations from Transformers for Biomedical Text Mining (BioBERT)) in extracting acupoint-related location relations and assess the impact of pretraining and fine-tuning on GPT's performance. We utilized the World Health Organization Standard Acupuncture Point Locations in the Western Pacific Region (WHO Standard) as our corpus, which consists of descriptions of 361 acupoints. Five types of relations ('direction_of,' 'distance_of,' 'part_of,' 'near_acupoint,' and 'located_near') (n= 3,174) between acupoints were annotated. Five models were compared: BioBERT, LSTM, pre-trained GPT-3.5, fine-tuned GPT-3.5, as well as pre-trained GPT-4. Performance metrics included micro-average exact match precision, recall, and F1 scores. Our results demonstrate that fine-tuned GPT-3.5 consistently outperformed other models in F1 scores across all relation types. Overall, it achieved the highest micro-average F1 score of 0.92. This study underscores the effectiveness of LLMs like GPT in extracting relations related to acupoint locations, with implications for accurately modeling acupuncture knowledge and promoting standard implementation in acupuncture training and practice. The findings also contribute to advancing informatics applications in traditional and complementary medicine, showcasing the potential of LLMs in natural language processing.
Authors: Haotian Luo
Abstract: Ensuring alignment with human preferences is a crucial characteristic of large language models (LLMs). Presently, the primary alignment methods, RLHF and DPO, require extensive human annotation, which is expensive despite their efficacy. The significant expenses associated with current alignment techniques motivate researchers to investigate the development of annotation-free alignment training methods. In pursuit of improved alignment without relying on external annotation, we introduce Latent Distance Guided Alignment Training (LD-Align). This approach seeks to align the model with a high-quality supervised fine-tune dataset using guidance from a latent space. The latent space is generated through sample reconstruction, akin to auto-encoding. Consequently, we utilize the distance between sample pairs in the latent space to guide DPO-based alignment training. Extensive experimentation and evaluation show the efficacy of our proposed method in achieving notable alignment.
Authors: Linan Yue, Qi Liu, Lili Zhao, Li Wang, Weibo Gao, Yanqing An
Abstract: With the development of legal intelligence, Criminal Court View Generation has attracted much attention as a crucial task of legal intelligence, which aims to generate concise and coherent texts that summarize case facts and provide explanations for verdicts. Existing researches explore the key information in case facts to yield the court views. Most of them employ a coarse-grained approach that partitions the facts into broad segments (e.g., verdict-related sentences) to make predictions. However, this approach fails to capture the complex details present in the case facts, such as various criminal elements and legal events. To this end, in this paper, we propose an Event Grounded Generation (EGG) method for criminal court view generation with cooperative (Large) Language Models, which introduces the fine-grained event information into the generation. Specifically, we first design a LLMs-based extraction method that can extract events in case facts without massive annotated events. Then, we incorporate the extracted events into court view generation by merging case facts and events. Besides, considering the computational burden posed by the use of LLMs in the extraction phase of EGG, we propose a LLMs-free EGG method that can eliminate the requirement for event extraction using LLMs in the inference phase. Extensive experimental results on a real-world dataset clearly validate the effectiveness of our proposed method.
Authors: Kuo-Yu Liao, Cheng-Shang Chang, Y. -W. Peter Hong
Abstract: Recent advances in Large Language Models (LLMs) have demonstrated the emergence of capabilities (learned skills) when the number of system parameters and the size of training data surpass certain thresholds. The exact mechanisms behind such phenomena are not fully understood and remain a topic of active research. Inspired by the skill-text bipartite graph model presented in [1] for modeling semantic language, we develop a mathematical theory to explain the emergence of learned skills, taking the learning (or training) process into account. Our approach models the learning process for skills in the skill-text bipartite graph as an iterative decoding process in Low-Density Parity Check (LDPC) codes and Irregular Repetition Slotted ALOHA (IRSA). Using density evolution analysis, we demonstrate the emergence of learned skills when the ratio of the size of training texts to the number of skills exceeds a certain threshold. Our analysis also yields a scaling law for testing errors relative to the size of training texts. Upon completion of the training, we propose a method for semantic compression and discuss its application in semantic communication.
Authors: Michael Dalvean
Abstract: Word complexity is defined in a number of different ways. Psycholinguistic, morphological and lexical proxies are often used. Human ratings are also used. The problem here is that these proxies do not measure complexity directly, and human ratings are susceptible to subjective bias. In this study we contend that some form of 'latent complexity' can be approximated by using samples of simple and complex words. We use a sample of 'simple' words from primary school picture books and a sample of 'complex' words from high school and academic settings. In order to analyse the differences between these classes, we look at the letter positional probabilities (LPPs). We find strong statistical associations between several LPPs and complexity. For example, simple words are significantly (p<.001) more likely to start with w, b, s, h, g, k, j, t, y or f, while complex words are significantly (p<.001) more likely to start with i, a, e, r, v, u or d. We find similar strong associations for subsequent letter positions, with 84 letter-position variables in the first 6 positions being significant at the p<.001 level. We then use LPPs as variables in creating a classifier which can classify the two classes with an 83% accuracy. We test these findings using a second data set, with 66 LPPs significant (p<.001) in the first 6 positions common to both datasets. We use these 66 variables to create a classifier that is able to classify a third dataset with an accuracy of 70%. Finally, we create a fourth sample by combining the extreme high and low scoring words generated by three classifiers built on the first three separate datasets and use this sample to build a classifier which has an accuracy of 97%. We use this to score the four levels of English word groups from an ESL program.
Authors: Chi Tran, Huong Le Thanh
Abstract: Large Language Models (LLMs) and Multimodal Large language models (MLLMs) have taken the world by storm with impressive abilities in complex reasoning and linguistic comprehension. Meanwhile there are plethora of works related to Vietnamese Large Language Models, the lack of high-quality resources in multimodality limits the progress of Vietnamese MLLMs. In this paper, we pioneer in address this by introducing LaVy, a state-of-the-art Vietnamese MLLM, and we also introduce LaVy-Bench benchmark designated for evaluating MLLMs's understanding on Vietnamese visual language tasks. All code and model weights are public at https://github.com/baochi0212/LaVy
Authors: Kenan Jiang, Xuehai He, Ruize Xu, Xin Eric Wang
Abstract: Contrastive Language-Image Pretraining (CLIP) has demonstrated great zero-shot performance for matching images and text. However, it is still challenging to adapt vision-lanaguage pretrained models like CLIP to compositional image and text matching -- a more challenging image and text matching task requiring the model understanding of compositional word concepts and visual components. Towards better compositional generalization in zero-shot image and text matching, in this paper, we study the problem from a causal perspective: the erroneous semantics of individual entities are essentially confounders that cause the matching failure. Therefore, we propose a novel \textbf{\textit{training-free}} compositional CLIP model (ComCLIP). ComCLIP disentangles input images into subjects, objects, and action sub-images and composes CLIP's vision encoder and text encoder to perform evolving matching over compositional text embedding and sub-image embeddings. In this way, ComCLIP can mitigate spurious correlations introduced by the pretrained CLIP models and dynamically evaluate the importance of each component. Experiments on four compositional image-text matching datasets: SVO, ComVG, Winoground, and VL-checklist, and two general image-text retrieval datasets: Flick30K, and MSCOCO demonstrate the effectiveness of our plug-and-play method, which boosts the \textbf{\textit{zero-shot}} inference ability of CLIP, SLIP, and BLIP2 even without further training or fine-tuning. Our codes can be found at https://github.com/eric-ai-lab/ComCLIP.
Authors: Yutaro Yamada, Yingtian Tang, Yoyo Zhang, Ilker Yildirim
Abstract: Large-scale vision-language models such as CLIP have shown impressive performance on zero-shot image classification and image-to-text retrieval. However, such performance does not realize in tasks that require a finer-grained correspondence between vision and language, such as Visual Question Answering (VQA). As a potential cause of the difficulty of applying these models to VQA and similar tasks, we report an interesting phenomenon of vision-language models, which we call the Concept Association Bias (CAB). We find that models with CAB tend to treat input as a bag of concepts and attempt to fill in the other missing concept crossmodally, leading to an unexpected zero-shot prediction. We demonstrate CAB by showing that CLIP's zero-shot classification performance greatly suffers when there is a strong concept association between an object (e.g. eggplant) and an attribute (e.g. color purple). We also show that the strength of CAB predicts the performance on VQA. We observe that CAB is prevalent in vision-language models trained with contrastive losses, even when autoregressive losses are jointly employed. However, a model that solely relies on autoregressive loss seems to exhibit minimal or no signs of CAB.
Authors: Yi Zhong, Chen Zhang, Xule Liu, Chenxi Sun, Weishan Deng, Haifeng Hu, Zhongqian Sun
Abstract: While Current TTS systems perform well in synthesizing high-quality speech, producing highly expressive speech remains a challenge. Emphasis, as a critical factor in determining the expressiveness of speech, has attracted more attention nowadays. Previous works usually enhance the emphasis by adding intermediate features, but they can not guarantee the overall expressiveness of the speech. To resolve this matter, we propose Emphatic Expressive TTS (EE-TTS), which leverages multi-level linguistic information from syntax and semantics. EE-TTS contains an emphasis predictor that can identify appropriate emphasis positions from text and a conditioned acoustic model to synthesize expressive speech with emphasis and linguistic information. Experimental results indicate that EE-TTS outperforms baseline with MOS improvements of 0.49 and 0.67 in expressiveness and naturalness. EE-TTS also shows strong generalization across different datasets according to AB test results.
Authors: Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Yarin Gal, Nicolas Papernot, Ross Anderson
Abstract: Stable Diffusion revolutionised image creation from descriptive text. GPT-2, GPT-3(.5) and GPT-4 demonstrated astonishing performance across a variety of language tasks. ChatGPT introduced such language models to the general public. It is now clear that large language models (LLMs) are here to stay, and will bring about drastic change in the whole ecosystem of online text and images. In this paper we consider what the future might hold. What will happen to GPT-{n} once LLMs contribute much of the language found online? We find that use of model-generated content in training causes irreversible defects in the resulting models, where tails of the original content distribution disappear. We refer to this effect as Model Collapse and show that it can occur in Variational Autoencoders, Gaussian Mixture Models and LLMs. We build theoretical intuition behind the phenomenon and portray its ubiquity amongst all learned generative models. We demonstrate that it has to be taken seriously if we are to sustain the benefits of training from large-scale data scraped from the web. Indeed, the value of data collected about genuine human interactions with systems will be increasingly valuable in the presence of content generated by LLMs in data crawled from the Internet.
Authors: Madhur Panwar, Kabir Ahuja, Navin Goyal
Abstract: In-context learning (ICL) is one of the surprising and useful features of large language models and subject of intense research. Recently, stylized meta-learning-like ICL setups have been devised that train transformers on sequences of input-output pairs $(x, f(x))$. The function $f$ comes from a function class and generalization is checked by evaluating on sequences generated from unseen functions from the same class. One of the main discoveries in this line of research has been that for several function classes, such as linear regression, transformers successfully generalize to new functions in the class. However, the inductive biases of these models resulting in this behavior are not clearly understood. A model with unlimited training data and compute is a Bayesian predictor: it learns the pretraining distribution. In this paper we empirically examine how far this Bayesian perspective can help us understand ICL. To this end, we generalize the previous meta-ICL setup to hierarchical meta-ICL setup which involve unions of multiple task families. We instantiate this setup on a diverse range of linear and nonlinear function families and find that transformers can do ICL in this setting as well. Where Bayesian inference is tractable, we find evidence that high-capacity transformers mimic the Bayesian predictor. The Bayesian perspective provides insights into the inductive bias of ICL and how transformers perform a particular task when they are trained on multiple tasks. We also find that transformers can learn to generalize to new function classes that were not seen during pretraining. This involves deviation from the Bayesian predictor. We examine these deviations in more depth offering new insights and hypotheses.
Authors: Wenqi Fan, Zihuai Zhao, Jiatong Li, Yunqing Liu, Xiaowei Mei, Yiqi Wang, Zhen Wen, Fei Wang, Xiangyu Zhao, Jiliang Tang, Qing Li
Abstract: With the prosperity of e-commerce and web applications, Recommender Systems (RecSys) have become an important component of our daily life, providing personalized suggestions that cater to user preferences. While Deep Neural Networks (DNNs) have made significant advancements in enhancing recommender systems by modeling user-item interactions and incorporating textual side information, DNN-based methods still face limitations, such as difficulties in understanding users' interests and capturing textual side information, inabilities in generalizing to various recommendation scenarios and reasoning on their predictions, etc. Meanwhile, the emergence of Large Language Models (LLMs), such as ChatGPT and GPT4, has revolutionized the fields of Natural Language Processing (NLP) and Artificial Intelligence (AI), due to their remarkable abilities in fundamental responsibilities of language understanding and generation, as well as impressive generalization and reasoning capabilities. As a result, recent studies have attempted to harness the power of LLMs to enhance recommender systems. Given the rapid evolution of this research direction in recommender systems, there is a pressing need for a systematic overview that summarizes existing LLM-empowered recommender systems, to provide researchers in relevant fields with an in-depth understanding. Therefore, in this paper, we conduct a comprehensive review of LLM-empowered recommender systems from various aspects including Pre-training, Fine-tuning, and Prompting. More specifically, we first introduce representative methods to harness the power of LLMs (as a feature encoder) for learning representations of users and items. Then, we review recent techniques of LLMs for enhancing recommender systems from three paradigms, namely pre-training, fine-tuning, and prompting. Finally, we comprehensively discuss future directions in this emerging field.
Authors: Chengrun Yang, Xuezhi Wang, Yifeng Lu, Hanxiao Liu, Quoc V. Le, Denny Zhou, Xinyun Chen
Abstract: Optimization is ubiquitous. While derivative-based algorithms have been powerful tools for various problems, the absence of gradient imposes challenges on many real-world applications. In this work, we propose Optimization by PROmpting (OPRO), a simple and effective approach to leverage large language models (LLMs) as optimizers, where the optimization task is described in natural language. In each optimization step, the LLM generates new solutions from the prompt that contains previously generated solutions with their values, then the new solutions are evaluated and added to the prompt for the next optimization step. We first showcase OPRO on linear regression and traveling salesman problems, then move on to our main application in prompt optimization, where the goal is to find instructions that maximize the task accuracy. With a variety of LLMs, we demonstrate that the best prompts optimized by OPRO outperform human-designed prompts by up to 8% on GSM8K, and by up to 50% on Big-Bench Hard tasks. Code at https://github.com/google-deepmind/opro.
Authors: Junzhe Jiang, Shang Qu, Mingyue Cheng, Qi Liu, Zhiding Liu, Hao Zhang, Rujiao Zhang, Kai Zhang, Rui Li, Jiatong Li, Min Gao
Abstract: Recommender systems are indispensable in the realm of online applications, and sequential recommendation has enjoyed considerable prevalence due to its capacity to encapsulate the dynamic shifts in user interests. However, previous sequential modeling methods still have limitations in capturing contextual information. The primary reason is the lack of understanding of domain-specific knowledge and item-related textual content. Fortunately, the emergence of powerful language models has unlocked the potential to incorporate extensive world knowledge into recommendation algorithms, enabling them to go beyond simple item attributes and truly understand the world surrounding user preferences. To achieve this, we propose LANCER, which leverages the semantic understanding capabilities of pre-trained language models to generate personalized recommendations. Our approach bridges the gap between language models and recommender systems, resulting in more human-like recommendations. We demonstrate the effectiveness of our approach through a series of experiments conducted on multiple benchmark datasets, showing promising results and providing valuable insights into the influence of our model on sequential recommendation tasks. Furthermore, our experimental codes are publicly available at https://github.com/Gnimixy/lancer.
Authors: Zijian Yang, Wei Zhou, Ralf Schl\"uter, Hermann Ney
Abstract: Internal language model (ILM) subtraction has been widely applied to improve the performance of the RNN-Transducer with external language model (LM) fusion for speech recognition. In this work, we show that sequence discriminative training has a strong correlation with ILM subtraction from both theoretical and empirical points of view. Theoretically, we derive that the global optimum of maximum mutual information (MMI) training shares a similar formula as ILM subtraction. Empirically, we show that ILM subtraction and sequence discriminative training achieve similar effects across a wide range of experiments on Librispeech, including both MMI and minimum Bayes risk (MBR) criteria, as well as neural transducers and LMs of both full and limited context. The benefit of ILM subtraction also becomes much smaller after sequence discriminative training. We also provide an in-depth study to show that sequence discriminative training has a minimal effect on the commonly used zero-encoder ILM estimation, but a joint effect on both encoder and prediction + joint network for posterior probability reshaping including both ILM and blank suppression.
Authors: Yaxin Zhu, Hamed Zamani
Abstract: This paper focuses on the task of Extreme Multi-Label Classification (XMC) whose goal is to predict multiple labels for each instance from an extremely large label space. While existing research has primarily focused on fully supervised XMC, real-world scenarios often lack supervision signals, highlighting the importance of zero-shot settings. Given the large label space, utilizing in-context learning approaches is not trivial. We address this issue by introducing In-Context Extreme Multilabel Learning (ICXML), a two-stage framework that cuts down the search space by generating a set of candidate labels through incontext learning and then reranks them. Extensive experiments suggest that ICXML advances the state of the art on two diverse public benchmarks.
Authors: Derek Tam, Mohit Bansal, Colin Raffel
Abstract: Model merging aims to cheaply combine individual task-specific models into a single multitask model. In this work, we view past merging methods as leveraging different notions of a ''task parameter subspace'' in which models are matched before being merged. We connect the task parameter subspace of a given model to its loss landscape and formalize how this approach to model merging can be seen as solving a linear system of equations. While past work has generally been limited to linear systems that have a closed-form solution, we consider using the conjugate gradient method to find a solution. We show that using the conjugate gradient method can outperform closed-form solutions, enables merging via linear systems that are otherwise intractable to solve, and flexibly allows choosing from a wide variety of initializations and estimates for the ''task parameter subspace''. We ultimately demonstrate that our merging framework called ''Matching Models in their Task Parameter Subspace'' (MaTS) achieves state-of-the-art results in multitask and intermediate-task model merging. We release all of the code and checkpoints used in our work at https://github.com/r-three/mats.
Authors: Jian Gu, Aldeida Aleti, Chunyang Chen, Hongyu Zhang
Abstract: Large Language Models (LLMs) have found widespread adoption in software engineering, particularly in code generation tasks. However, updating these models with new knowledge can be prohibitively expensive, yet it is essential for maximizing their utility. In this paper, we propose a novel and effective model editing approach, \textsc{MENT}, to patch LLMs in coding tasks. \textsc{MENT} is effective, efficient, and reliable. It can correct a neural model by patching 1 or 2 neurons. As the pioneer work on neuron-level model editing of generative models, we formalize the editing process and introduce the involved concepts. Besides, we also introduce new measures to evaluate its generalization ability, and build a benchmark for further study. Our approach is evaluated on three coding tasks, including API-seq recommendation, line-level code generation, and pseudocode-to-code transaction. The experimental results show that the proposed approach outperforms the state of the arts by a significant margin in both effectiveness and efficiency measures. In addition, we demonstrate the usages of \textsc{MENT} for LLM reasoning in software engineering. By editing LLM knowledge, the directly or indirectly dependent behaviors of API invocation in the chain-of-thought will change accordingly. It explained the significance of repairing LLMs.
Authors: Xin Ding, Xiaoyu Liu, Zhijun Tu, Yun Zhang, Wei Li, Jie Hu, Hanting Chen, Yehui Tang, Zhiwei Xiong, Baoqun Yin, Yunhe Wang
Abstract: Post-training quantization (PTQ) has played a key role in compressing large language models (LLMs) with ultra-low costs. However, existing PTQ methods only focus on handling the outliers within one layer or one block, which ignores the dependency of blocks and leads to severe performance degradation in low-bit settings. In this paper, we propose CBQ, a cross-block reconstruction-based PTQ method for LLMs. CBQ employs a cross-block dependency using a homologous reconstruction scheme, establishing long-range dependencies across multiple blocks to minimize error accumulation. Furthermore, CBQ incorporates a coarse-to-fine preprocessing (CFP) strategy for suppressing weight and activation outliers, coupled with an adaptive LoRA-Rounding technique for precise weight quantization. These innovations enable CBQ to not only handle extreme outliers effectively but also improve overall quantization accuracy. Extensive experiments show that CBQ achieves superior low-bit quantization (W4A4, W4A8, W2A16) and outperforms existing state-of-the-art methods across various LLMs and datasets. Notably, CBQ quantizes the 4-bit LLAMA1-65B model within only 4.3 hours on a single GPU, achieving a commendable tradeoff between performance and quantization efficiency.
Authors: Xiaoxu Xu, Yitian Yuan, Qiudan Zhang, Wenhui Wu, Zequn Jie, Lin Ma, Xu Wang
Abstract: Learning to ground natural language queries to target objects or regions in 3D point clouds is quite essential for 3D scene understanding. Nevertheless, existing 3D visual grounding approaches require a substantial number of bounding box annotations for text queries, which is time-consuming and labor-intensive to obtain. In this paper, we propose \textbf{3D-VLA}, a weakly supervised approach for \textbf{3D} visual grounding based on \textbf{V}isual \textbf{L}inguistic \textbf{A}lignment. Our 3D-VLA exploits the superior ability of current large-scale vision-language models (VLMs) on aligning the semantics between texts and 2D images, as well as the naturally existing correspondences between 2D images and 3D point clouds, and thus implicitly constructs correspondences between texts and 3D point clouds with no need for fine-grained box annotations in the training procedure. During the inference stage, the learned text-3D correspondence will help us ground the text queries to the 3D target objects even without 2D images. To the best of our knowledge, this is the first work to investigate 3D visual grounding in a weakly supervised manner by involving large scale vision-language models, and extensive experiments on ReferIt3D and ScanRefer datasets demonstrate that our 3D-VLA achieves comparable and even superior results over the fully supervised methods.
Authors: Yuanyi Wang, Haifeng Sun, Jingyu Wang, Qi Qi, Shaoling Sun, Jianxin Liao
Abstract: Entity alignment (EA), a pivotal process in integrating multi-source Knowledge Graphs (KGs), seeks to identify equivalent entity pairs across these graphs. Most existing approaches regard EA as a graph representation learning task, concentrating on enhancing graph encoders. However, the decoding process in EA - essential for effective operation and alignment accuracy - has received limited attention and remains tailored to specific datasets and model architectures, necessitating both entity and additional explicit relation embeddings. This specificity limits its applicability, particularly in GNN-based models. To address this gap, we introduce a novel, generalized, and efficient decoding approach for EA, relying solely on entity embeddings. Our method optimizes the decoding process by minimizing Dirichlet energy, leading to the gradient flow within the graph, to maximize graph homophily. The discretization of the gradient flow produces a fast and scalable approach, termed Triple Feature Propagation (TFP). TFP innovatively generalizes adjacency matrices to multi-views matrices:entity-to-entity, entity-to-relation, relation-to-entity, and relation-to-triple. The gradient flow through generalized matrices enables TFP to harness the multi-view structural information of KGs. Rigorous experimentation on diverse public datasets demonstrates that our approach significantly enhances various EA methods. Notably, the approach achieves these advancements with less than 6 seconds of additional computational time, establishing a new benchmark in efficiency and adaptability for future EA methods.
Authors: Fangkai Jiao, Chengwei Qin, Zhengyuan Liu, Nancy F. Chen, Shafiq Joty
Abstract: Large Language Models (LLMs) have demonstrated significant potential in handling complex reasoning tasks through step-by-step rationale generation. However, recent studies have raised concerns regarding the hallucination and flaws in their reasoning process. Substantial efforts are being made to improve the reliability and faithfulness of the generated rationales. Some approaches model reasoning as planning, while others focus on annotating for process supervision. Nevertheless, the planning-based search process often results in high latency due to the frequent assessment of intermediate reasoning states and the extensive exploration space. Additionally, supervising the reasoning process with human annotation is costly and challenging to scale for LLM training. To address these issues, in this paper, we propose a framework to learn planning-based reasoning through Direct Preference Optimization (DPO) on collected trajectories, which are ranked according to synthesized process rewards. Our results on challenging logical reasoning benchmarks demonstrate the effectiveness of our learning framework, showing that our 7B model can surpass the strong counterparts like GPT-3.5-Turbo.
Authors: Wasu Top Piriyakulkij, Kevin Ellis
Abstract: We build a computational model of how humans actively infer hidden rules by doing experiments. The basic principles behind the model is that, even if the rule is deterministic, the learner considers a broader space of fuzzy probabilistic rules, which it represents in natural language, and updates its hypotheses online after each experiment according to approximately Bayesian principles. In the same framework we also model experiment design according to information-theoretic criteria. We find that the combination of these three principles -- explicit hypotheses, probabilistic rules, and online updates -- can explain human performance on a Zendo-style task, and that removing any of these components leaves the model unable to account for the data.
Authors: Milan Bhan, Jean-Noel Vittaut, Nicolas Chesneau, Marie-Jeanne Lesot
Abstract: Incorporating natural language rationales in the prompt and In-Context Learning (ICL) has led to a significant improvement of Large Language Models (LLMs) performance. However, rationales currently require human-annotation or the use of auxiliary proxy models to target promising samples or generate high-quality rationales. In this work, we propose Self-AMPLIFY to generate automatically rationales from post hoc explanation methods applied to Small Language Models (SLMs) to improve their own performance. Self-AMPLIFY is a 3-step method that targets samples, generates rationales and builds a final prompt to leverage ICL. Self-AMPLIFY performance is evaluated on two SLMs and two datasets requiring reasoning abilities: these experiments show that Self-AMPLIFY achieves good results against competitors. Self-AMPLIFY is the first method to apply post hoc explanation methods to SLM to generate rationales to improve their own performance in a fully automated manner.
Authors: Junzhe Zhang, Huixuan Zhang, Xunjian Yin, Xiaojun Wan
Abstract: News image captioning requires model to generate an informative caption rich in entities, with the news image and the associated news article. Though Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in addressing various vision-language tasks, our research finds that current MLLMs still bear limitations in handling entity information on news image captioning task. Besides, while MLLMs have the ability to process long inputs, generating high-quality news image captions still requires a trade-off between sufficiency and conciseness of textual input information. To explore the potential of MLLMs and address problems we discovered, we propose : an Entity-Aware Multimodal Alignment based approach for news image captioning. Our approach first aligns the MLLM through Balance Training Strategy with two extra alignment tasks: Entity-Aware Sentence Selection task and Entity Selection task, together with News Image Captioning task, to enhance its capability in handling multimodal entity information. The aligned MLLM will utilizes the additional entity-related information it explicitly extract to supplement its textual input while generating news image captions. Our approach achieves better results than all previous models in CIDEr score on GoodNews dataset (72.33 -> 88.39) and NYTimes800k dataset (70.83 -> 85.61).
Authors: Indraneil Paul, Goran Glava\v{s}, Iryna Gurevych
Abstract: Code understanding and generation have fast become some of the most popular applications of language models (LMs). Nonetheless, research on multilingual aspects of Code-LMs (i.e., LMs for code generation) such as cross-lingual transfer between different programming languages, language-specific data augmentation, and post-hoc LM adaptation, alongside exploitation of data sources other than the original textual content, has been much sparser than for their natural language counterparts. In particular, most mainstream Code-LMs have been pre-trained on source code files alone. In this work, we investigate the prospect of leveraging readily available compiler intermediate representations (IR) - shared across programming languages - to improve the multilingual capabilities of Code-LMs and facilitate cross-lingual transfer. To this end, we first compile SLTrans, a parallel dataset consisting of nearly 4M self-contained source code files coupled with respective intermediate representations. Next, starting from various base Code-LMs (ranging in size from 1.1B to 7.3B parameters), we carry out continued causal language modelling training on SLTrans, forcing the Code-LMs to (1) learn the IR language and (2) align the IR constructs with respective constructs of various programming languages. Our resulting models, dubbed IRCoder, display sizeable and consistent gains across a wide variety of code generation tasks and metrics, including prompt robustness, multilingual code completion, code understanding, and instruction following.
Authors: Thang M. Pham, Peijie Chen, Tin Nguyen, Seunghyun Yoon, Trung Bui, Anh Totti Nguyen
Abstract: CLIP-based classifiers rely on the prompt containing a {class name} that is known to the text encoder. Therefore, they perform poorly on new classes or the classes whose names rarely appear on the Internet (e.g., scientific names of birds). For fine-grained classification, we propose PEEB - an explainable and editable classifier to (1) express the class name into a set of text descriptors that describe the visual parts of that class; and (2) match the embeddings of the detected parts to their textual descriptors in each class to compute a logit score for classification. In a zero-shot setting where the class names are unknown, PEEB outperforms CLIP by a huge margin (~10x in top-1 accuracy). Compared to part-based classifiers, PEEB is not only the state-of-the-art (SOTA) on the supervised-learning setting (88.80% and 92.20% accuracy on CUB-200 and Dogs-120, respectively) but also the first to enable users to edit the text descriptors to form a new classifier without any re-training. Compared to concept bottleneck models, PEEB is also the SOTA in both zero-shot and supervised-learning settings.
Authors: Yifan Li, Hangyu Guo, Kun Zhou, Wayne Xin Zhao, Ji-Rong Wen
Abstract: In this paper, we study the harmlessness alignment problem of multimodal large language models (MLLMs). We conduct a systematic empirical analysis of the harmlessness performance of representative MLLMs and reveal that the image input poses the alignment vulnerability of MLLMs. Inspired by this, we propose a novel jailbreak method named HADES, which hides and amplifies the harmfulness of the malicious intent within the text input, using meticulously crafted images. Experimental results show that HADES can effectively jailbreak existing MLLMs, which achieves an average Attack Success Rate (ASR) of 90.26% for LLaVA-1.5 and 71.60% for Gemini Pro Vision. Our code and data will be publicly released.
Authors: Abhishek Ghose, Emma Thuong Nguyen
Abstract: Active learning (AL) techniques aim to maximally utilize a labeling budget by iteratively selecting instances that are most likely to improve prediction accuracy. However, their benefit compared to random sampling has not been consistent across various setups, e.g., different datasets, classifiers. In this empirical study, we examine how a combination of different factors might obscure any gains from an AL technique. Focusing on text classification, we rigorously evaluate AL techniques over around 1000 experiments that vary wrt the dataset, batch size, text representation and the classifier. We show that AL is only effective in a narrow set of circumstances. We also address the problem of using metrics that are better aligned with real world expectations. The impact of this study is in its insights for a practitioner: (a) the choice of text representation and classifier is as important as that of an AL technique, (b) choice of the right metric is critical in assessment of the latter, and, finally, (c) reported AL results must be holistically interpreted, accounting for variables other than just the query strategy.
Authors: Akash Ghosh, Arkadeep Acharya, Sriparna Saha, Vinija Jain, Aman Chadha
Abstract: The advent of Large Language Models (LLMs) has significantly reshaped the trajectory of the AI revolution. Nevertheless, these LLMs exhibit a notable limitation, as they are primarily adept at processing textual information. To address this constraint, researchers have endeavored to integrate visual capabilities with LLMs, resulting in the emergence of Vision-Language Models (VLMs). These advanced models are instrumental in tackling more intricate tasks such as image captioning and visual question answering. In our comprehensive survey paper, we delve into the key advancements within the realm of VLMs. Our classification organizes VLMs into three distinct categories: models dedicated to vision-language understanding, models that process multimodal inputs to generate unimodal (textual) outputs and models that both accept and produce multimodal inputs and outputs.This classification is based on their respective capabilities and functionalities in processing and generating various modalities of data.We meticulously dissect each model, offering an extensive analysis of its foundational architecture, training data sources, as well as its strengths and limitations wherever possible, providing readers with a comprehensive understanding of its essential components. We also analyzed the performance of VLMs in various benchmark datasets. By doing so, we aim to offer a nuanced understanding of the diverse landscape of VLMs. Additionally, we underscore potential avenues for future research in this dynamic domain, anticipating further breakthroughs and advancements.
Authors: Kevin Zhang, Luka Chkhetiani, Francis McCann Ramirez, Yash Khare, Andrea Vanzo, Michael Liang, Sergio Ramirez Martin, Gabriel Oexle, Ruben Bousbib, Taufiquzzaman Peyash, Michael Nguyen, Dillon Pulliam, Domenic Donato
Abstract: This paper presents Conformer-1, an end-to-end Automatic Speech Recognition (ASR) model trained on an extensive dataset of 570k hours of speech audio data, 91% of which was acquired from publicly available sources. To achieve this, we perform Noisy Student Training after generating pseudo-labels for the unlabeled public data using a strong Conformer RNN-T baseline model. The addition of these pseudo-labeled data results in remarkable improvements in relative Word Error Rate (WER) by 11.5% and 24.3% for our asynchronous and realtime models, respectively. Additionally, the model is more robust to background noise owing to the addition of these data. The results obtained in this study demonstrate that the incorporation of pseudo-labeled publicly available data is a highly effective strategy for improving ASR accuracy and noise robustness.
Authors: Jonathan D. Chang, Wenhao Zhan, Owen Oertell, Kiant\'e Brantley, Dipendra Misra, Jason D. Lee, Wen Sun
Abstract: Reinforcement Learning (RL) from Human Preference-based feedback is a popular paradigm for fine-tuning generative models, which has produced impressive models such as GPT-4 and Claude3 Opus. This framework often consists of two steps: learning a reward model from an offline preference dataset followed by running online RL to optimize the learned reward model. In this work, leveraging the idea of reset, we propose a new RLHF algorithm with provable guarantees. Motivated by the fact that offline preference dataset provides informative states (i.e., data that is preferred by the labelers), our new algorithm, Dataset Reset Policy Optimization (DR-PO), integrates the existing offline preference dataset into the online policy training procedure via dataset reset: it directly resets the policy optimizer to the states in the offline dataset, instead of always starting from the initial state distribution. In theory, we show that DR-PO learns to perform at least as good as any policy that is covered by the offline dataset under general function approximation with finite sample complexity. In experiments, we demonstrate that on both the TL;DR summarization and the Anthropic Helpful Harmful (HH) dataset, the generation from DR-PO is better than that from Proximal Policy Optimization (PPO) and Direction Preference Optimization (DPO), under the metric of GPT4 win-rate. Code for this work can be found at https://github.com/Cornell-RL/drpo.