Authors: Y. Shen, Z. Ji, J. Lin, K. R. Koedginer
Abstract: Protecting Personally Identifiable Information (PII), such as names, is a critical requirement in learning technologies to safeguard student and teacher privacy and maintain trust. Accurate PII detection is an essential step toward anonymizing sensitive information while preserving the utility of educational data. Motivated by recent advancements in artificial intelligence, our study investigates the GPT-4o-mini model as a cost-effective and efficient solution for PII detection tasks. We explore both prompting and fine-tuning approaches and compare GPT-4o-mini's performance against established frameworks, including Microsoft Presidio and Azure AI Language. Our evaluation on two public datasets, CRAPII and TSCC, demonstrates that the fine-tuned GPT-4o-mini model achieves superior performance, with a recall of 0.9589 on CRAPII. Additionally, fine-tuned GPT-4o-mini significantly improves precision scores (a threefold increase) while reducing computational costs to nearly one-tenth of those associated with Azure AI Language. Furthermore, our bias analysis reveals that the fine-tuned GPT-4o-mini model consistently delivers accurate results across diverse cultural backgrounds and genders. The generalizability analysis using the TSCC dataset further highlights its robustness, achieving a recall of 0.9895 with minimal additional training data from TSCC. These results emphasize the potential of fine-tuned GPT-4o-mini as an accurate and cost-effective tool for PII detection in educational data. It offers robust privacy protection while preserving the data's utility for research and pedagogical analysis. Our code is available on GitHub: https://github.com/AnonJD/PrivacyAI
Authors: Yirong Zeng, Xiao Ding, Yuxian Wang, Weiwen Liu, Wu Ning, Yutai Hou, Xu Huang, Bing Qin, Ting Liu
Abstract: Augmenting large language models (LLMs) with external tools is a promising approach to enhance their capabilities. Effectively leveraging this potential for complex tasks hinges crucially on improving their ability to use tools. Synthesizing tool use data by simulating the real world is an effective approach. Nevertheless, our investigation reveals that training gains significantly decay as the scale of these data increases. The primary factor is the model's poor performance (a.k.a deficiency) in complex scenarios, which hinders learning from data using SFT. Driven by this objective, we propose an iterative reinforced fine-tuning strategy to continually guide the model to alleviate it. Specifically, we first identify deficiency-related data based on feedback from the policy model, then perform a Monte Carlo Tree Search to collect fine-grained preference pairs to pinpoint deficiencies. Subsequently, we update the policy model using preference optimization to align with ground truth and misalign with deficiencies. This process can be iterated. Moreover, before the iteration, we propose an easy-to-hard warm-up SFT strategy to facilitate learning from challenging data. The experiments demonstrate our models go beyond the same parametric models, outperforming many larger open-source and closed-source models. Additionally, it has achieved notable training gains in complex tool use scenarios.
Authors: Tuowei Wang, Xingyu Chen, Kun Li, Ting Cao, Ju Ren, Yaoxue Zhang
Abstract: The escalating demand for long-context applications has intensified the necessity of extending the LLM context windows. Despite recent fine-tuning approaches successfully expanding context lengths, their high memory footprints, especially for activations, present a critical practical limitation. Current parameter-efficient fine-tuning methods prioritize reducing parameter update overhead over addressing activation memory constraints. Similarly, existing sparsity mechanisms improve computational efficiency but overlook activation memory optimization due to the phenomenon of Shadowy Activation. In this paper, we propose LeMo, the first LLM fine-tuning system that explores and exploits a new token-level sparsity mechanism inherent in long-context scenarios, termed Contextual Token Sparsity. LeMo minimizes redundant token involvement by assessing the informativeness of token embeddings while preserving model accuracy. Specifically, LeMo introduces three key techniques: (1) Token Elimination, dynamically identifying and excluding redundant tokens across varying inputs and layers. (2) Pattern Prediction, utilizing well-trained predictors to approximate token sparsity patterns with minimal overhead. (3) Kernel Optimization, employing permutation-free and segment-based strategies to boost system performance. We implement LeMo as an end-to-end fine-tuning system compatible with various LLM architectures and other optimization techniques. Comprehensive evaluations demonstrate that LeMo reduces memory consumption by up to 1.93x and achieves up to 1.36x speedups, outperforming state-of-the-art fine-tuning systems.
Authors: Mohamed Bayan Kmainasi, Ali Ezzat Shahroor, Amani Al-Ghraibah
Abstract: Large Language Models (LLMs) have shown exceptional capabilities in Natural Language Processing (NLP) across diverse domains. However, their application in specialized tasks such as Legal Judgment Prediction (LJP) for low-resource languages like Arabic remains underexplored. In this work, we address this gap by developing an Arabic LJP dataset, collected and preprocessed from Saudi commercial court judgments. We benchmark state-of-the-art open-source LLMs, including LLaMA-3.2-3B and LLaMA-3.1-8B, under varying configurations such as zero-shot, one-shot, and fine-tuning using QLoRA. Additionally, we used a comprehensive evaluation framework combining quantitative metrics (BLEU and ROUGE) and qualitative assessments (Coherence, legal language, clarity). Our results demonstrate that fine-tuned smaller models achieve comparable performance to larger models in task-specific contexts while offering significant resource efficiency. Furthermore, we investigate the effects of prompt engineering and fine-tuning on model outputs, providing insights into performance variability and instruction sensitivity. By making the dataset, implementation code, and models publicly available, we establish a robust foundation for future research in Arabic legal NLP.
Authors: Tairan Fu, Javier Conde, Gonzalo Mart\'inez, Mar\'ia Grandury, Pedro Reviriego
Abstract: One of the most widely used methods to evaluate LLMs are Multiple Choice Question (MCQ) tests. MCQ benchmarks enable the testing of LLM knowledge on almost any topic at scale as the results can be processed automatically. To help the LLM answer, a few examples called few shots can be included in the prompt. Moreover, the LLM can be asked to answer the question directly with the selected option or to first provide the reasoning and then the selected answer, which is known as chain of thought. In addition to checking whether the selected answer is correct, the evaluation can look at the LLM-estimated probability of its response as an indication of the confidence of the LLM in the response. In this paper, we study how the LLM confidence in its answer depends on whether the model has been asked to answer directly or to provide the reasoning before answering. The results of the evaluation of questions on a wide range of topics in seven different models show that LLMs are more confident in their answers when they provide reasoning before the answer. This occurs regardless of whether the selected answer is correct. Our hypothesis is that this behavior is due to the reasoning that modifies the probability of the selected answer, as the LLM predicts the answer based on the input question and the reasoning that supports the selection made. Therefore, LLM estimated probabilities seem to have intrinsic limitations that should be understood in order to use them in evaluation procedures. Interestingly, the same behavior has been observed in humans, for whom explaining an answer increases confidence in its correctness.
Authors: Vahid Amiri, Mahmood Ahmadi
Abstract: Cryptocurrency is a digital currency that uses blockchain technology with secure encryption. Due to the decentralization of these currencies, traditional monetary systems and the capital market of each they, can influence a society. Therefore, due to the importance of the issue, the need to understand public opinion and analyze people's opinions in this regard increases. To understand the opinions and views of people about different topics, you can take help from social networks because they are a rich source of opinions. The Twitter social network is one of the main platforms where users discuss various topics, therefore, in the shortest time and with the lowest cost, the opinion of the community can be measured on this social network. Twitter Sentiment Analysis (TSA) is a field that analyzes the sentiment expressed in tweets. Considering that most of TSA's research efforts on cryptocurrencies are focused on English language, the purpose of this paper is to investigate the opinions of Iranian users on the Twitter social network about cryptocurrencies and provide the best model for classifying tweets based on sentiment. In the case of automatic analysis of tweets, managers and officials in the field of economy can gain knowledge from the general public's point of view about this issue and use the information obtained in order to properly manage this phenomenon. For this purpose, in this paper, in order to build emotion classification models, natural language processing techniques such as bag of words (BOW) and FastText for text vectorization and classical machine learning algorithms including KNN, SVM and Adaboost learning methods Deep including LSTM and BERT model were used for classification, and finally BERT linguistic model had the best accuracy with 83.50%.
Authors: Teodor-George Marchitan, Claudiu Creanga, Liviu P. Dinu
Abstract: This paper describes the approach of the Unibuc - NLP team in tackling the Coling 2025 GenAI Workshop, Task 1: Binary Multilingual Machine-Generated Text Detection. We explored both masked language models and causal models. For Subtask A, our best model achieved first-place out of 36 teams when looking at F1 Micro (Auxiliary Score) of 0.8333, and second-place when looking at F1 Macro (Main Score) of 0.8301
Authors: Nada Saadi, Tathagata Raha, Cl\'ement Christophe, Marco AF Pimentel, Ronnie Rajan, Praveen K Kanithi
Abstract: This paper investigates the challenges of developing large language models (LLMs) proficient in both multilingual understanding and medical knowledge. We demonstrate that simply translating medical data does not guarantee strong performance on clinical tasks in the target language. Our experiments reveal that the optimal language mix in training data varies significantly across different medical tasks. We find that larger models with carefully calibrated language ratios achieve superior performance on native-language clinical tasks. Furthermore, our results suggest that relying solely on fine-tuning may not be the most effective approach for incorporating new language knowledge into LLMs. Instead, data and computationally intensive pretraining methods may still be necessary to achieve optimal performance in multilingual medical settings. These findings provide valuable guidance for building effective and inclusive medical AI systems for diverse linguistic communities.
Authors: Reham Omar, Omij Mangukiya, Essam Mansour
Abstract: Dialogue benchmarks are crucial in training and evaluating chatbots engaging in domain-specific conversations. Knowledge graphs (KGs) represent semantically rich and well-organized data spanning various domains, such as DBLP, DBpedia, and YAGO. Traditionally, dialogue benchmarks have been manually created from documents, neglecting the potential of KGs in automating this process. Some question-answering benchmarks are automatically generated using extensive preprocessing from KGs, but they do not support dialogue generation. This paper introduces Chatty-Gen, a novel multi-stage retrieval-augmented generation platform for automatically generating high-quality dialogue benchmarks tailored to a specific domain using a KG. Chatty-Gen decomposes the generation process into manageable stages and uses assertion rules for automatic validation between stages. Our approach enables control over intermediate results to prevent time-consuming restarts due to hallucinations. It also reduces reliance on costly and more powerful commercial LLMs. Chatty-Gen eliminates upfront processing of the entire KG using efficient query-based retrieval to find representative subgraphs based on the dialogue context. Our experiments with several real and large KGs demonstrate that Chatty-Gen significantly outperforms state-of-the-art systems and ensures consistent model and system performance across multiple LLMs of diverse capabilities, such as GPT-4o, Gemini 1.5, Llama 3, and Mistral.
Authors: Zuhong Liu, Charles-Elie Simon, Fabien Caspani
Abstract: Retrieval-Augmented Generation (RAG) has proven effective in open-domain question answering. However, the chunking process, which is essential to this pipeline, often receives insufficient attention relative to retrieval and synthesis components. This study emphasizes the critical role of chunking in improving the performance of both dense passage retrieval and the end-to-end RAG pipeline. We then introduce the Logits-Guided Multi-Granular Chunker (LGMGC), a novel framework that splits long documents into contextualized, self-contained chunks of varied granularity. Our experimental results, evaluated on two benchmark datasets, demonstrate that LGMGC not only improves the retrieval step but also outperforms existing chunking methods when integrated into a RAG pipeline.
Authors: Belu Ticona, Fernando Carranza, Viviana Cotik
Abstract: Argentina has a diverse, yet little-known, Indigenous language heritage. Most of these languages are at risk of disappearing, resulting in a significant loss of world heritage and cultural knowledge. Currently, no unified information on speakers and computational tools is available for these languages. In this work, we present a systematization of the Indigenous languages spoken in Argentina, along with national demographic data on the country's Indigenous population. The languages are classified into seven families: Mapuche, Tup\'i-Guaran\'i, Guaycur\'u, Quechua, Mataco-Mataguaya, Aymara, and Chon. We also provide an introductory survey of the computational resources available for these languages, whether or not they are specifically developed for Argentine varieties.
Authors: Zengyi Gao, Yukun Cao, Hairu Wang, Ao Ke, Yuan Feng, Xike Xie, S Kevin Zhou
Abstract: To mitigate the hallucination and knowledge deficiency in large language models (LLMs), Knowledge Graph (KG)-based Retrieval-Augmented Generation (RAG) has shown promising potential by utilizing KGs as external resource to enhance LLMs reasoning.However, existing KG-RAG approaches struggle with a trade-off between flexibility and retrieval quality.Modular methods prioritize flexibility by avoiding the use of KG-fine-tuned models during retrieval, leading to fixed retrieval strategies and suboptimal retrieval quality.Conversely, coupled methods embed KG information within models to improve retrieval quality, but at the expense of flexibility.In this paper, we propose a novel flexible modular KG-RAG framework, termed FRAG, which synergizes the advantages of both approaches.FRAG estimates the hop range of reasoning paths based solely on the query and classify it as either simple or complex.To match the complexity of the query, tailored pipelines are applied to ensure efficient and accurate reasoning path retrieval, thus fostering the final reasoning process.By using the query text instead of the KG to infer the structural information of reasoning paths and employing adaptable retrieval strategies, FRAG improves retrieval quality while maintaining flexibility.Moreover, FRAG does not require extra LLMs fine-tuning or calls, significantly boosting efficiency and conserving resources.Extensive experiments show that FRAG achieves state-of-the-art performance with high efficiency and low resource consumption.
Authors: Chen Zhang, Xinyi Dai, Yaxiong Wu, Qu Yang, Yasheng Wang, Ruiming Tang, Yong Liu
Abstract: Multi-turn interaction in the dialogue system research refers to a system's ability to maintain context across multiple dialogue turns, enabling it to generate coherent and contextually relevant responses. Recent advancements in large language models (LLMs) have significantly expanded the scope of multi-turn interaction, moving beyond chatbots to enable more dynamic agentic interactions with users or environments. In this paper, we provide a focused review of the multi-turn capabilities of LLMs, which are critical for a wide range of downstream applications, including conversational search and recommendation, consultation services, and interactive tutoring. This survey explores four key aspects: (1) the core model capabilities that contribute to effective multi-turn interaction, (2) how multi-turn interaction is evaluated in current practice, (3) the general algorithms used to enhance multi-turn interaction, and (4) potential future directions for research in this field.
Authors: Yeonseok Jeong, Minsoo Kim, Seung-won Hwang, Byung-Hak Kim
Abstract: Large Language Models (LLMs) have demonstrated near-human performance in summarization tasks based on traditional metrics such as ROUGE and BERTScore. However, these metrics do not adequately capture critical aspects of summarization quality, such as factual accuracy, particularly for long narratives (>100K tokens). Recent advances, such as LLM-as-a-Judge, address the limitations of metrics based on lexical similarity but still exhibit factual inconsistencies, especially in understanding character relationships and states. In this work, we introduce NarrativeFactScore, a novel "Agent-as-a-Judge" framework for evaluating and refining summaries. By leveraging a Character Knowledge Graph (CKG) extracted from input and generated summaries, NarrativeFactScore assesses the factual consistency and provides actionable guidance for refinement, such as identifying missing or erroneous facts. We demonstrate the effectiveness of NarrativeFactScore through a detailed workflow illustration and extensive validation on widely adopted benchmarks, achieving superior performance compared to competitive methods. Our results highlight the potential of agent-driven evaluation systems to improve the factual reliability of LLM-generated summaries.
Authors: Qiang Liu, Xinlong Chen, Yue Ding, Shizhen Xu, Shu Wu, Liang Wang
Abstract: Hallucination has emerged as a significant barrier to the effective application of Large Language Models (LLMs). In this work, we introduce a novel Attention-Guided SElf-Reflection (AGSER) approach for zero-shot hallucination detection in LLMs. The AGSER method utilizes attention contributions to categorize the input query into attentive and non-attentive queries. Each query is then processed separately through the LLMs, allowing us to compute consistency scores between the generated responses and the original answer. The difference between the two consistency scores serves as a hallucination estimator. In addition to its efficacy in detecting hallucinations, AGSER notably reduces computational complexity, requiring only three passes through the LLM and utilizing two sets of tokens. We have conducted extensive experiments with four widely-used LLMs across three different hallucination benchmarks, demonstrating that our approach significantly outperforms existing methods in zero-shot hallucination detection.
Authors: Bidit Sadhukhan, Swami Punyeshwarananda
Abstract: Sanskrit, one of humanity's most ancient languages, has a vast collection of books and manuscripts on diverse topics that have been accumulated over millennia. However, its digital content (audio and text), which is vital for the training of AI systems, is profoundly limited. Furthermore, its intricate linguistics make it hard to develop robust NLP tools for wider accessibility. Given these constraints, we have developed an automatic speech recognition model for Sanskrit by employing transfer learning mechanism on OpenAI's Whisper model. After carefully optimising the hyper-parameters, we obtained promising results with our transfer-learned model achieving a word error rate of 15.42% on Vaksancayah dataset. An online demo of our model is made available for the use of public and to evaluate its performance firsthand thereby paving the way for improved accessibility and technological support for Sanskrit learning in the modern era.
Authors: Paul R\"ottger, Giuseppe Attanasio, Felix Friedrich, Janis Goldzycher, Alicia Parrish, Rishabh Bhardwaj, Chiara Di Bonaventura, Roman Eng, Gaia El Khoury Geagea, Sujata Goswami, Jieun Han, Dirk Hovy, Seogyeong Jeong, Paloma Jereti\v{c}, Flor Miriam Plaza-del-Arco, Donya Rooein, Patrick Schramowski, Anastassia Shaitarova, Xudong Shen, Richard Willats, Andrea Zugarini, Bertie Vidgen
Abstract: Vision-language models (VLMs), which process image and text inputs, are increasingly integrated into chat assistants and other consumer AI applications. Without proper safeguards, however, VLMs may give harmful advice (e.g. how to self-harm) or encourage unsafe behaviours (e.g. to consume drugs). Despite these clear hazards, little work so far has evaluated VLM safety and the novel risks created by multimodal inputs. To address this gap, we introduce MSTS, a Multimodal Safety Test Suite for VLMs. MSTS comprises 400 test prompts across 40 fine-grained hazard categories. Each test prompt consists of a text and an image that only in combination reveal their full unsafe meaning. With MSTS, we find clear safety issues in several open VLMs. We also find some VLMs to be safe by accident, meaning that they are safe because they fail to understand even simple test prompts. We translate MSTS into ten languages, showing non-English prompts to increase the rate of unsafe model responses. We also show models to be safer when tested with text only rather than multimodal prompts. Finally, we explore the automation of VLM safety assessments, finding even the best safety classifiers to be lacking.
Authors: Patrick Krauss, Achim Schilling
Abstract: Deep learning methods have been increasingly applied to computational linguistics to uncover patterns in text data. This study investigates author-specific word class distributions using part-of-speech (POS) tagging and bigram analysis. By leveraging deep neural networks, we classify literary authors based on POS tag vectors and bigram frequency matrices derived from their works. We employ fully connected and convolutional neural network architectures to explore the efficacy of unigram and bigram-based representations. Our results demonstrate that while unigram features achieve moderate classification accuracy, bigram-based models significantly improve performance, suggesting that sequential word class patterns are more distinctive of authorial style. Multi-dimensional scaling (MDS) visualizations reveal meaningful clustering of authors' works, supporting the hypothesis that stylistic nuances can be captured through computational methods. These findings highlight the potential of deep learning and linguistic feature analysis for author profiling and literary studies.
Authors: Latofat Bobojonova, Arofat Akhundjanova, Phil Ostheimer, Sophie Fellenz
Abstract: This paper advances NLP research for the low-resource Uzbek language by evaluating two previously untested monolingual Uzbek BERT models on the part-of-speech (POS) tagging task and introducing the first publicly available UPOS-tagged benchmark dataset for Uzbek. Our fine-tuned models achieve 91% average accuracy, outperforming the baseline multi-lingual BERT as well as the rule-based tagger. Notably, these models capture intermediate POS changes through affixes and demonstrate context sensitivity, unlike existing rule-based taggers.
Authors: Lucen Zhong, Zhengxiao Du, Xiaohan Zhang, Haiyi Hu, Jie Tang
Abstract: Enhancing large language models (LLMs) with real-time APIs can help generate more accurate and up-to-date responses. However, evaluating the function calling abilities of LLMs in real-world scenarios remains under-explored due to the complexity of data collection and evaluation. In this work, we introduce ComplexFuncBench, a benchmark for complex function calling across five real-world scenarios. Compared to existing benchmarks, ComplexFuncBench encompasses multi-step and constrained function calling, which requires long-parameter filing, parameter value reasoning, and 128k long context. Additionally, we propose an automatic framework, ComplexEval, for quantitatively evaluating complex function calling tasks. Through comprehensive experiments, we demonstrate the deficiencies of state-of-the-art LLMs in function calling and suggest future directions for optimizing these capabilities. The data and code are available at \url{https://github.com/THUDM/ComplexFuncBench}.
Authors: Tomasz Limisiewicz, David Mare\v{c}ek, Tom\'a\v{s} Musil
Abstract: Mitigation of biases, such as language models' reliance on gender stereotypes, is a crucial endeavor required for the creation of reliable and useful language technology. The crucial aspect of debiasing is to ensure that the models preserve their versatile capabilities, including their ability to solve language tasks and equitably represent various genders. To address this issue, we introduce a streamlined Dual Dabiasing Algorithm through Model Adaptation (2DAMA). Novel Dual Debiasing enables robust reduction of stereotypical bias while preserving desired factual gender information encoded by language models. We show that 2DAMA effectively reduces gender bias in English and is one of the first approaches facilitating the mitigation of stereotypical tendencies in translation. The proposed method's key advantage is the preservation of factual gender cues, which are useful in a wide range of natural language processing tasks.
Authors: Vera Pavlova
Abstract: This study examines the use of Natural Language Processing (NLP) technology within the Islamic domain, focusing on developing an Islamic neural retrieval model. By leveraging the robust XLM-R model, the research employs a language reduction technique to create a lightweight bilingual large language model (LLM). Our approach for domain adaptation addresses the unique challenges faced in the Islamic domain, where substantial in-domain corpora exist only in Arabic while limited in other languages, including English. The work utilizes a multi-stage training process for retrieval models, incorporating large retrieval datasets, such as MS MARCO, and smaller, in-domain datasets to improve retrieval performance. Additionally, we have curated an in-domain retrieval dataset in English by employing data augmentation techniques and involving a reliable Islamic source. This approach enhances the domain-specific dataset for retrieval, leading to further performance gains. The findings suggest that combining domain adaptation and a multi-stage training method for the bilingual Islamic neural retrieval model enables it to outperform monolingual models on downstream retrieval tasks.
Authors: Suvodip Dey, Yi-Jyun Sun, Gokhan Tur, Dilek Hakkani-Tur
Abstract: Recent LLMs have enabled significant advancements for conversational agents. However, they are also well-known to hallucinate, i.e., they often produce responses that seem plausible but are not factually correct. On the other hand, users tend to over-rely on LLM-based AI agents; they accept the AI's suggestion even when it is wrong. Adding good friction, such as explanations or getting user confirmations, has been proposed as a mitigation in AI-supported decision-making systems. In this paper, we propose an accountability model for LLM-based task-oriented dialogue agents to address user overreliance via friction turns in cases of model uncertainty and errors associated with dialogue state tracking (DST). The accountability model is an augmented LLM with an additional accountability head, which functions as a binary classifier to predict the slots of the dialogue states. We perform our experiments with three backbone LLMs (Llama, Mistral, Gemma) on two established task-oriented datasets (MultiWOZ and Snips). Our empirical findings demonstrate that this approach not only enables reliable estimation of AI agent errors but also guides the LLM decoder in generating more accurate actions. We observe around 3% absolute improvement in joint goal accuracy by incorporating accountability heads in modern LLMs for the MultiWOZ dataset. We also show that this method enables the agent to self-correct its actions, further boosting its performance by 3%. Finally, we discuss the application of accountability modeling to prevent user overreliance by introducing friction.
Authors: Andrick Adhikari, Sanchari Das, Rinku Dewri
Abstract: Natural Language Processing (NLP) is an essential subset of artificial intelligence. It has become effective in several domains, such as healthcare, finance, and media, to identify perceptions, opinions, and misuse, among others. Privacy is no exception, and initiatives have been taken to address the challenges of usable privacy notifications to users with the help of NLP. To this aid, we conduct a literature review by analyzing 109 papers at the intersection of NLP and privacy policies. First, we provide a brief introduction to privacy policies and discuss various facets of associated problems, which necessitate the application of NLP to elevate the current state of privacy notices and disclosures to users. Subsequently, we a) provide an overview of the implementation and effectiveness of NLP approaches for better privacy policy communication; b) identify the methodologies that can be further enhanced to provide robust privacy policies; and c) identify the gaps in the current state-of-the-art research. Our systematic analysis reveals that several research papers focus on annotating and classifying privacy texts for analysis but need to adequately dwell on other aspects of NLP applications, such as summarization. More specifically, ample research opportunities exist in this domain, covering aspects such as corpus generation, summarization vectors, contextualized word embedding, identification of privacy-relevant statement categories, fine-grained classification, and domain-specific model tuning.
Authors: Pit Neitemeier, Bj\"orn Deiseroth, Constantin Eichenberg, Lukas Balles
Abstract: Tokenization is a fundamental step in natural language processing, breaking text into units that computational models can process. While learned subword tokenizers have become the de-facto standard, they present challenges such as large vocabularies, limited adaptability to new domains or languages, and sensitivity to spelling errors and variations. To overcome these limitations, we investigate a hierarchical architecture for autoregressive language modelling that combines character-level and word-level processing. It employs a lightweight character-level encoder to convert character sequences into word embeddings, which are then processed by a word-level backbone model and decoded back into characters via a compact character-level decoder. This method retains the sequence compression benefits of word-level tokenization without relying on a rigid, predefined vocabulary. We demonstrate, at scales up to 7 billion parameters, that hierarchical transformers match the downstream task performance of subword-tokenizer-based models while exhibiting significantly greater robustness to input perturbations. Additionally, during continued pretraining on an out-of-domain language, our model trains almost twice as fast, achieves superior performance on the target language, and retains more of its previously learned knowledge. Hierarchical transformers pave the way for NLP systems that are more robust, flexible, and generalizable across languages and domains.
Authors: Suvodip Dey, Maunendra Sankar Desarkar
Abstract: The standard language modeling (LM) loss by itself has been shown to be inadequate for effective dialogue modeling. As a result, various training approaches, such as auxiliary loss functions and leveraging human feedback, are being adopted to enrich open-domain dialogue systems. One such auxiliary loss function is Bag-of-Words (BoW) loss, defined as the cross-entropy loss for predicting all the words/tokens of the next utterance. In this work, we propose a novel auxiliary loss named Bag-of-Keywords (BoK) loss to capture the central thought of the response through keyword prediction and leverage it to enhance the generation of meaningful and interpretable responses in open-domain dialogue systems. BoK loss upgrades the BoW loss by predicting only the keywords or critical words/tokens of the next utterance, intending to estimate the core idea rather than the entire response. We incorporate BoK loss in both encoder-decoder (T5) and decoder-only (DialoGPT) architecture and train the models to minimize the weighted sum of BoK and LM (BoK-LM) loss. We perform our experiments on two popular open-domain dialogue datasets, DailyDialog and Persona-Chat. We show that the inclusion of BoK loss improves the dialogue generation of backbone models while also enabling post-hoc interpretability. We also study the effectiveness of BoK-LM loss as a reference-free metric and observe comparable performance to the state-of-the-art metrics on various dialogue evaluation datasets.
Authors: Andrey Labunets, Nishit V. Pandya, Ashish Hooda, Xiaohan Fu, Earlence Fernandes
Abstract: We surface a new threat to closed-weight Large Language Models (LLMs) that enables an attacker to compute optimization-based prompt injections. Specifically, we characterize how an attacker can leverage the loss-like information returned from the remote fine-tuning interface to guide the search for adversarial prompts. The fine-tuning interface is hosted by an LLM vendor and allows developers to fine-tune LLMs for their tasks, thus providing utility, but also exposes enough information for an attacker to compute adversarial prompts. Through an experimental analysis, we characterize the loss-like values returned by the Gemini fine-tuning API and demonstrate that they provide a useful signal for discrete optimization of adversarial prompts using a greedy search algorithm. Using the PurpleLlama prompt injection benchmark, we demonstrate attack success rates between 65% and 82% on Google's Gemini family of LLMs. These attacks exploit the classic utility-security tradeoff - the fine-tuning interface provides a useful feature for developers but also exposes the LLMs to powerful attacks.
Authors: Soham Roy, Mitul Goswami, Nisharg Nargund, Suneeta Mohanty, Prasant Kumar Pattnaik
Abstract: This study introduces a system leveraging Large Language Models (LLMs) to extract text and enhance user interaction with PDF documents via a conversational interface. Utilizing Retrieval-Augmented Generation (RAG), the system provides informative responses to user inquiries while highlighting relevant passages within the PDF. Upon user upload, the system processes the PDF, employing sentence embeddings to create a document-specific vector store. This vector store enables efficient retrieval of pertinent sections in response to user queries. The LLM then engages in a conversational exchange, using the retrieved information to extract text and generate comprehensive, contextually aware answers. While our approach demonstrates competitive ROUGE values compared to existing state-of-the-art techniques for text extraction and summarization, we acknowledge that further qualitative evaluation is necessary to fully assess its effectiveness in real-world applications. The proposed system gives competitive ROUGE values as compared to existing state-of-the-art techniques for text extraction and summarization, thus offering a valuable tool for researchers, students, and anyone seeking to efficiently extract knowledge and gain insights from documents through an intuitive question-answering interface.
Authors: Maxwell J. Yin, Dingyi Jiang, Yongbing Chen, Boyu Wang, Charles Ling
Abstract: Chain-of-Thought (CoT) reasoning in smaller language models is a challenging natural language process problem yet highly desirable in many real-life applications. Existing CoT knowledge distillation methods often suffer from overly conservative memorization in smaller LLMs, leading to low generalization confidence. As fully preserving the CoT ability of teacher model is impossible, we hypothesize that adversarial CoT fine-tuning is crucial for developing smaller LLM with robust CoT generalization. To this end, we propose \textit{PRompt-Assisted Domain-Adversarial fine-tuning} (PRADA), a principled fine-tuning framework that integrates diverse CoT domains. Specifically, PRADA pioneers two CoT improvements in smaller LLM: (1) Recovering the domain-invariant feature insight which typically lost during distillation with domain adversarial fine-tuning; (2) Enhancing the domain adaptability of CoT prompt engineering by employing domain-adversarial approaches. We theoretically demonstrate the effectiveness of our approach and empirically show that it significantly outperforms the state of the arts in a wide range of tasks. Moreover, our empirical findings reveal that the smaller LLM, when leveraging PRADA, aligns closely with domain knowledge, thereby improving the explainability of our approach.
Authors: Samuel Soo, Wesley Teng, Chandrasekaran Balaganesh
Abstract: Effective and reliable control over large language model (LLM) behavior is a significant challenge. While activation steering methods, which add steering vectors to a model's hidden states, are a promising approach, existing techniques often lack precision and interpretability in how they influence model outputs. We introduce Feature Guided Activation Additions (FGAA), a novel activation steering method that leverages insights from Contrastive Activation Addition (CAA) and Sparse Autoencoder-Targeted Steering (SAE-TS). By operating in the latent space of a Sparse Autoencoder (SAE) and employing optimization techniques to select desired SAE features, FGAA constructs precise steering vectors that provide better steering effects while maintaining coherence of steered model outputs. In this regard, evaluations on Gemma-2-2B and Gemma-2-9B models across various steering tasks demonstrate that FGAA outperforms existing steering methods of CAA, SAE decoder steering, and SAE-TS. Our results also highlight important trade-offs between steering scale and general model capabilities that are consistent across all tested steering methods.
Authors: Qingcheng Zeng, Guanhong Liu, Zhaoqian Xue, Diego Ford, Rob Voigt, Loni Hagen, Lingyao Li
Abstract: On July 13, 2024, at the Trump rally in Pennsylvania, someone attempted to assassinate Republican Presidential Candidate Donald Trump. This attempt sparked a large-scale discussion on social media. We collected posts from X (formerly known as Twitter) one week before and after the assassination attempt and aimed to model the short-term effects of such a ``shock'' on public opinions and discussion topics. Specifically, our study addresses three key questions: first, we investigate how public sentiment toward Donald Trump shifts over time and across regions (RQ1) and examine whether the assassination attempt itself significantly affects public attitudes, independent of the existing political alignments (RQ2). Finally, we explore the major themes in online conversations before and after the crisis, illustrating how discussion topics evolved in response to this politically charged event (RQ3). By integrating large language model-based sentiment analysis, difference-in-differences modeling, and topic modeling techniques, we find that following the attempt the public response was broadly sympathetic to Trump rather than polarizing, despite baseline ideological and regional disparities.
Authors: Yuefan Cao, Chengyue Gong, Xiaoyu Li, Yingyu Liang, Zhizhou Sha, Zhenmei Shi, Zhao Song
Abstract: Text-to-video generation models have made impressive progress, but they still struggle with generating videos with complex features. This limitation often arises from the inability of the text encoder to produce accurate embeddings, which hinders the video generation model. In this work, we propose a novel approach to overcome this challenge by selecting the optimal text embedding through interpolation in the embedding space. We demonstrate that this method enables the video generation model to produce the desired videos. Additionally, we introduce a simple algorithm using perpendicular foot embeddings and cosine similarity to identify the optimal interpolation embedding. Our findings highlight the importance of accurate text embeddings and offer a pathway for improving text-to-video generation performance.
Authors: Jinyuan Feng, Zhiqiang Pu, Tianyi Hu, Dongmin Li, Xiaolin Ai, Huimu Wang
Abstract: Building mixture-of-experts (MoE) architecture for Low-rank adaptation (LoRA) is emerging as a potential direction in parameter-efficient fine-tuning (PEFT) for its modular design and remarkable performance. However, simply stacking the number of experts cannot guarantee significant improvement. In this work, we first conduct qualitative analysis to indicate that experts collapse to similar representations in vanilla MoE, limiting the capacity of modular design and computational efficiency. Ulteriorly, Our analysis reveals that the performance of previous MoE variants maybe limited by a lack of diversity among experts. Motivated by these findings, we propose Orthogonal Mixture-of-Experts (OMoE), a resource-efficient MoE variant that trains experts in an orthogonal manner to promote diversity. In OMoE, a Gram-Schmidt process is leveraged to enforce that the experts' representations lie within the Stiefel manifold. By applying orthogonal constraints directly to the architecture, OMoE keeps the learning objective unchanged, without compromising optimality. Our method is simple and alleviates memory bottlenecks, as it incurs minimal experts compared to vanilla MoE models. Experiments on diverse commonsense reasoning benchmarks demonstrate that OMoE can consistently achieve stable and efficient performance improvement when compared with the state-of-the-art methods while significantly reducing the number of required experts.
Authors: Kazuma Onishi, Katsuhiko Hayashi
Abstract: Extreme multi-label learning (XML) is a task of assigning multiple labels from an extremely large set of labels to each data instance. Many current high-performance XML models are composed of a lot of hyperparameters, which complicates the tuning process. Additionally, the models themselves are adapted specifically to XML, which complicates their reimplementation. To remedy this problem, we propose a simple method based on ridge regression for XML. The proposed method not only has a closed-form solution but also is composed of a single hyperparameter. Since there are no precedents on applying ridge regression to XML, this paper verified the performance of the method by using various XML benchmark datasets. Furthermore, we enhanced the prediction of low-frequency labels in XML, which hold informative content. This prediction is essential yet challenging because of the limited amount of data. Here, we employed a simple frequency-based weighting. This approach greatly simplifies the process compared with existing techniques. Experimental results revealed that it can achieve levels of performance comparable to, or even exceeding, those of models with numerous hyperparameters. Additionally, we found that the frequency-based weighting significantly improved the predictive performance for low-frequency labels, while requiring almost no changes in implementation. The source code for the proposed method is available on github at https://github.com/cars1015/XML-ridge.
Authors: Wenqi Fan, Yi Zhou, Shijie Wang, Yuyao Yan, Hui Liu, Qian Zhao, Le Song, Qing Li
Abstract: Considering the significance of proteins, computational protein science has always been a critical scientific field, dedicated to revealing knowledge and developing applications within the protein sequence-structure-function paradigm. In the last few decades, Artificial Intelligence (AI) has made significant impacts in computational protein science, leading to notable successes in specific protein modeling tasks. However, those previous AI models still meet limitations, such as the difficulty in comprehending the semantics of protein sequences, and the inability to generalize across a wide range of protein modeling tasks. Recently, LLMs have emerged as a milestone in AI due to their unprecedented language processing & generalization capability. They can promote comprehensive progress in fields rather than solving individual tasks. As a result, researchers have actively introduced LLM techniques in computational protein science, developing protein Language Models (pLMs) that skillfully grasp the foundational knowledge of proteins and can be effectively generalized to solve a diversity of sequence-structure-function reasoning problems. While witnessing prosperous developments, it's necessary to present a systematic overview of computational protein science empowered by LLM techniques. First, we summarize existing pLMs into categories based on their mastered protein knowledge, i.e., underlying sequence patterns, explicit structural and functional information, and external scientific languages. Second, we introduce the utilization and adaptation of pLMs, highlighting their remarkable achievements in promoting protein structure prediction, protein function prediction, and protein design studies. Then, we describe the practical application of pLMs in antibody design, enzyme design, and drug discovery. Finally, we specifically discuss the promising future directions in this fast-growing field.
Authors: Zhenzhen Zhuang, Jiandong Chen, Hongfeng Xu, Yuwen Jiang, Jialiang Lin
Abstract: Large language models (LLMs) have significantly impacted human society, influencing various domains. Among them, academia is not simply a domain affected by LLMs, but it is also the pivotal force in the development of LLMs. In academic publications, this phenomenon is represented during the incorporation of LLMs into the peer review mechanism for reviewing manuscripts. We proposed the concept of automated scholarly paper review (ASPR) in our previous paper. As the incorporation grows, it now enters the coexistence phase of ASPR and peer review, which is described in that paper. LLMs hold transformative potential for the full-scale implementation of ASPR, but they also pose new issues and challenges that need to be addressed. In this survey paper, we aim to provide a holistic view of ASPR in the era of LLMs. We begin with a survey to find out which LLMs are used to conduct ASPR. Then, we review what ASPR-related technological bottlenecks have been solved with the incorporation of LLM technology. After that, we move on to explore new methods, new datasets, new source code, and new online systems that come with LLMs for ASPR. Furthermore, we summarize the performance and issues of LLMs in ASPR, and investigate the attitudes and reactions of publishers and academia to ASPR. Lastly, we discuss the challenges associated with the development of LLMs for ASPR. We hope this survey can serve as an inspirational reference for the researchers and promote the progress of ASPR for its actual implementation.
Authors: Henry Watkins, Robert Gray, Adam Julius, Yee-Haur Mah, Walter H. L. Pinaya, Paul Wright, Ashwani Jha, Holger Engleitner, Jorge Cardoso, Sebastien Ourselin, Geraint Rees, Rolf Jaeger, Parashkev Nachev
Abstract: Radiological reports typically summarize the content and interpretation of imaging studies in unstructured form that precludes quantitative analysis. This limits the monitoring of radiological services to throughput undifferentiated by content, impeding specific, targeted operational optimization. Here we present Neuradicon, a natural language processing (NLP) framework for quantitative analysis of neuroradiological reports. Our framework is a hybrid of rule-based and artificial intelligence models to represent neurological reports in succinct, quantitative form optimally suited to operational guidance. We demonstrate the application of Neuradicon to operational phenotyping of a corpus of 336,569 reports, and report excellent generalizability across time and two independent healthcare institutions.
Authors: Vinu Sankar Sadasivan, Aounon Kumar, Sriram Balasubramanian, Wenxiao Wang, Soheil Feizi
Abstract: Large Language Models (LLMs) perform impressively well in various applications. However, the potential for misuse of these models in activities such as plagiarism, generating fake news, and spamming has raised concern about their responsible use. Consequently, the reliable detection of AI-generated text has become a critical area of research. AI text detectors have shown to be effective under their specific settings. In this paper, we stress-test the robustness of these AI text detectors in the presence of an attacker. We introduce recursive paraphrasing attack to stress test a wide range of detection schemes, including the ones using the watermarking as well as neural network-based detectors, zero shot classifiers, and retrieval-based detectors. Our experiments conducted on passages, each approximately 300 tokens long, reveal the varying sensitivities of these detectors to our attacks. Our findings indicate that while our recursive paraphrasing method can significantly reduce detection rates, it only slightly degrades text quality in many cases, highlighting potential vulnerabilities in current detection systems in the presence of an attacker. Additionally, we investigate the susceptibility of watermarked LLMs to spoofing attacks aimed at misclassifying human-written text as AI-generated. We demonstrate that an attacker can infer hidden AI text signatures without white-box access to the detection method, potentially leading to reputational risks for LLM developers. Finally, we provide a theoretical framework connecting the AUROC of the best possible detector to the Total Variation distance between human and AI text distributions. This analysis offers insights into the fundamental challenges of reliable detection as language models continue to advance. Our code is publicly available at https://github.com/vinusankars/Reliability-of-AI-text-detectors.
URLs: https://github.com/vinusankars/Reliability-of-AI-text-detectors.
Authors: Delong Liu, Haiwen Li, Zhicheng Zhao, Yuan Dong
Abstract: The goal of Text-to-Image Person Retrieval (TIPR) is to retrieve specific person images according to the given textual descriptions. A primary challenge in this task is bridging the substantial representational gap between visual and textual modalities. The prevailing methods map texts and images into unified embedding space for matching, while the intricate semantic correspondences between texts and images are still not effectively constructed. To address this issue, we propose a novel TIPR framework to build fine-grained interactions and alignment between person images and the corresponding texts. Specifically, via fine-tuning the Contrastive Language-Image Pre-training (CLIP) model, a visual-textual dual encoder is firstly constructed, to preliminarily align the image and text features. Secondly, a Text-guided Image Restoration (TIR) auxiliary task is proposed to map abstract textual entities to specific image regions, improving the alignment between local textual and visual embeddings. Additionally, a cross-modal triplet loss is presented to handle hard samples, and further enhance the model's discriminability for minor differences. Moreover, a pruning-based text data augmentation approach is proposed to enhance focus on essential elements in descriptions, thereby avoiding excessive model attention to less significant information. The experimental results show our proposed method outperforms state-of-the-art methods on three popular benchmark datasets, and the code will be made publicly available at https://github.com/Delong-liu-bupt/SEN.
Authors: Qiming Bao, Gael Gendron, Alex Yuxuan Peng, Wanjun Zhong, Neset Tan, Yang Chen, Michael Witbrock, Jiamou Liu
Abstract: Large language models (LLMs), such as LLaMA, Alpaca, Vicuna, GPT-3.5 and GPT-4, have advanced the performance of AI systems on various natural language processing tasks to human-like levels. However, their generalisation and robustness when performing logical reasoning has not been sufficiently assessed. To comprehensively evaluate this ability, we develop three new logical reasoning datasets named "ReClor-plus", "LogiQA-plus" and "LogiQAv2-plus" that extend standard logical reasoning datasets to evaluate the robustness of the LLM's reasoning. For each, we create three subsets: the first with randomly shuffled options, the second with the correct choices replaced by "none of the other options is correct", and the third with a combination of shuffling and substitution. Experiments on these datasets show that these simple augmentations greatly hinder the models' performance. Despite their high performance on the original publicly available datasets, we find that all models perform poorly on these newly constructed datasets. We also demonstrate that introducing task variations into the training set can markedly improve the model's performance on both the original and our developed datasets. Finally, we show that applying logic-driven data augmentation for fine-tuning and prompting can enhance generalisation in both discriminative and generative models, offering a path to improving their robustness for tasks involving logical reasoning. Source code and data are made publicly available at https://github.com/Strong-AI-Lab/Logical-and-abstract-reasoning.
URLs: https://github.com/Strong-AI-Lab/Logical-and-abstract-reasoning.
Authors: Yinhong Liu, Han Zhou, Zhijiang Guo, Ehsan Shareghi, Ivan Vuli\'c, Anna Korhonen, Nigel Collier
Abstract: Large Language Models (LLMs) have demonstrated promising capabilities as automatic evaluators in assessing the quality of generated natural language. However, LLMs still exhibit biases in evaluation and often struggle to generate coherent evaluations that align with human assessments. In this work, we first conduct a systematic study of the misalignment between LLM evaluators and human evaluation, revealing that existing calibration methods aimed at mitigating biases of LLMs are insufficient for effectively aligning LLM evaluators. Inspired by the use of preference data in RLHF, we formulate the evaluation as a ranking problem and introduce Pairwise-preference Search (PAIRS), an uncertainty-guided search-based rank aggregation method that employs LLMs to conduct pairwise comparisons locally and efficiently ranks candidate texts globally. PAIRS achieves state-of-the-art performance on representative evaluation tasks in long-form generations and demonstrates significant improvements over direct scoring. Furthermore, we provide insights into the role of pairwise preference in quantifying the transitivity of LLMs and demonstrate how PAIRS benefits from calibration using debiased pairwise evaluations.
Authors: Thibaut Thonet, Jos Rozen, Laurent Besacier
Abstract: Research on Large Language Models (LLMs) has recently witnessed an increasing interest in extending the models' context size to better capture dependencies within long documents. While benchmarks have been proposed to assess long-range abilities, existing efforts primarily considered generic tasks that are not necessarily aligned with real-world applications. In contrast, we propose a new benchmark for long-context LLMs focused on a practical meeting assistant scenario in which the long contexts consist of transcripts obtained by automatic speech recognition, presenting unique challenges for LLMs due to the inherent noisiness and oral nature of such data. Our benchmark, ELITR-Bench, augments the existing ELITR corpus by adding 271 manually crafted questions with their ground-truth answers, as well as noisy versions of meeting transcripts altered to target different Word Error Rate levels. Our experiments with 12 long-context LLMs on ELITR-Bench confirm the progress made across successive generations of both proprietary and open models, and point out their discrepancies in terms of robustness to transcript noise. We also provide a thorough analysis of our GPT-4-based evaluation, including insights from a crowdsourcing study. Our findings indicate that while GPT-4's scores align with human judges, its ability to distinguish beyond three score levels may be limited.
Authors: Wei Zou, Ziyuan Zhuang, Xiang Geng, Shujian Huang, Jia Liu, Jiajun Chen
Abstract: Paraphrase generation strives to generate high-quality and diverse expressions of a given text, a domain where diffusion models excel. Though SOTA diffusion generation 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 can facilitate only input segments to ensure paraphrase semantics, improving the results without external features. Experiments show that LDP better reconciles paraphrase generation quality and diversity than baselines. Further analysis shows that our method is also helpful to other similar text generations and domain adaptations
Authors: Jiaming Zhou, Shiwan Zhao, Hui Wang, Tian-Hao Zhang, Haoqin Sun, Xuechen Wang, Yong Qin
Abstract: The kNN-CTC model has proven to be effective for monolingual automatic speech recognition (ASR). However, its direct application to multilingual scenarios like code-switching, presents challenges. Although there is potential for performance improvement, a kNN-CTC model utilizing a single bilingual datastore can inadvertently introduce undesirable noise from the alternative language. To address this, we propose a novel kNN-CTC-based code-switching ASR (CS-ASR) framework that employs dual monolingual datastores and a gated datastore selection mechanism to reduce noise interference. Our method selects the appropriate datastore for decoding each frame, ensuring the injection of language-specific information into the ASR process. We apply this framework to cutting-edge CTC-based models, developing an advanced CS-ASR system. Extensive experiments demonstrate the remarkable effectiveness of our gated datastore mechanism in enhancing the performance of zero-shot Chinese-English CS-ASR.
Authors: Jiaxing Wu, Lin Ning, Luyang Liu, Harrison Lee, Neo Wu, Chao Wang, Sushant Prakash, Shawn O'Banion, Bradley Green, Jun Xie
Abstract: LLM-powered personalization agent systems employ Large Language Models (LLMs) to predict users' behavior from their past activities. However, their effectiveness often hinges on the ability to effectively leverage extensive, long user historical data due to its inherent noise and length of such data. Existing pretrained LLMs may generate summaries that are concise but lack the necessary context for downstream tasks, hindering their utility in personalization systems. To address these challenges, we introduce Reinforcement Learning from Prediction Feedback (RLPF). RLPF fine-tunes LLMs to generate concise, human-readable user summaries that are optimized for downstream task performance. By maximizing the usefulness of the generated summaries, RLPF effectively distills extensive user history data while preserving essential information for downstream tasks. Our empirical evaluation demonstrates significant improvements in both extrinsic downstream task utility and intrinsic summary quality, surpassing baseline methods by up to 22% on downstream task performance and achieving an up to 84.59% win rate on Factuality, Abstractiveness, and Readability. RLPF also achieves a remarkable 74% reduction in context length while improving performance on 16 out of 19 unseen tasks and/or datasets, showcasing its generalizability. This approach offers a promising solution for enhancing LLM personalization by effectively transforming long, noisy user histories into informative and human-readable representations.
Authors: Xindi Tong, Yujin Zhu, Shijian Fan, Liang Xu
Abstract: Long text summarization, gradually being essential for efficiently processing large volumes of information, stays challenging for Large Language Models (LLMs) such as GPT and LLaMA families because of the insufficient open-sourced training datasets and the high requirement of contextual details dealing. To address the issue, we design a novel zero-shot transfer learning framework, abbreviated as T3, to iteratively training a baseline LLM on an assistant task for the target task, where the former should own richer data resources and share structural or semantic similarity with the latter. In practice, T3 is approached to deal with the long text summarization task by utilizing question answering as the assistant task, and further validated its effectiveness on the BBC summary, NarraSum, FairytaleQA, and NLQuAD datasets, with up to nearly 14% improvement in ROUGE, 35% improvement in BLEU, and 16% improvement in Factscore compared to three baseline LLMs, demonstrating its potential for more assistant-target task combinations.
Authors: Zhen Zheng, Xin Ji, Taosong Fang, Fanghao Zhou, Chuanjie Liu, Gang Peng
Abstract: Large language models (LLMs) increasingly play an important role in a wide range of information processing and management tasks. Many of these tasks are performed in large batches or even offline, and the performance indictor for which is throughput. These tasks usually show the characteristic of prefix sharing, where different prompt input can partially show the common prefix. However, the existing LLM inference engines tend to optimize the streaming requests and show limitations of supporting the large batched tasks with the prefix sharing characteristic. The existing solutions use the LRU-based cache to reuse the KV context of common prefix between requests. The KV context that are about to be reused may prematurely evicted with the implicit cache management. Besides, the streaming oriented systems do not leverage the request-batch information and can not mix the decoding tokens with the prefill chunks to the best for the batched scenarios, and thus fails to saturate the GPU. We propose BatchLLM to address the above problems. BatchLLM explicitly identifies the common prefixes globally. The requests sharing the same prefix will be scheduled together to reuse the KV context the best. BatchLLM reorders the requests and schedules the requests with larger ratio of decoding first to better mix the decoding tokens with the latter prefill chunks, and applies memory-centric token batching to enlarge the token-batch sizes, which helps to increase the GPU utilization. Finally, BatchLLM optimizes the prefix-shared Attention kernel with horizontal fusion to reduce tail effect and kernel launch overhead. Extensive evaluation shows that BatchLLM outperforms vLLM and SGLang by 1.3$\times$ to 10.8$\times$ on a set of microbenchmarks and a typical industry workload under different hardware environments.
Authors: Wonjin Lee, Kyumin Kim, Sungjae Lee, Jihun Lee, Kwang In Kim
Abstract: Applying language models (LMs) to tables is challenging due to the inherent structural differences between two-dimensional tables and one-dimensional text for which the LMs were originally designed. Furthermore, when applying linearized tables to LMs, the maximum token lengths often imposed in self-attention calculations make it difficult to comprehensively understand the context spread across large tables. To address these challenges, we present PieTa (Piece of Table), a new framework for sub-table-based question answering (QA). PieTa operates through an iterative process of dividing tables into smaller windows, using LMs to select relevant cells within each window, and merging these cells into a sub-table. This multi-resolution approach captures dependencies across multiple rows and columns while avoiding the limitations caused by long context inputs. Instantiated as a simple iterative sub-table union algorithm, PieTa demonstrates improved performance over previous sub-table-based QA approaches.
Authors: Wang Liang
Abstract: Multilingual transfer ability, which reflects how well models fine-tuned on one source language can be applied to other languages, has been well studied in multilingual pre-trained models. However, the existence of such capability transfer between natural language and gene sequences/languages remains under explored.This study addresses this gap by drawing inspiration from the sentence-pair classification task used for evaluating sentence similarity in natural language. We constructed two analogous tasks: DNA-pair classification(DNA sequence similarity) and DNA-protein-pair classification(gene coding determination). These tasks were designed to validate the transferability of capabilities from natural language to gene sequences. Even a small-scale pre-trained model like GPT-2-small, which was pre-trained on English, achieved an accuracy of 78% on the DNA-pair classification task after being fine-tuned on English sentence-pair classification data(XTREME PAWS-X). While training a BERT model on multilingual text, the precision reached 89%. On the more complex DNA-protein-pair classification task, however, the model's output was barely distinguishable from random output.Experimental validation has confirmed that the transfer of capabilities from natural language to biological language is unequivocally present. Building on this foundation, we have also investigated the impact of model parameter scale and pre-training on this capability transfer. We provide recommendations for facilitating the transfer of capabilities from natural language to genetic language,as well as new approaches for conducting biological research based on this capability.This study offers an intriguing new perspective on exploring the relationship between natural language and genetic language.
Authors: Umer Butt, Stalin Veranasi, G\"unter Neumann
Abstract: As the Information Retrieval (IR) field increasingly recognizes the importance of inclusivity, addressing the needs of low-resource languages remains a significant challenge. This paper introduces the first large-scale Urdu IR dataset, created by translating the MS MARCO dataset through machine translation. We establish baseline results through zero-shot learning for IR in Urdu and subsequently apply the mMARCO multilingual IR methodology to this newly translated dataset. Our findings demonstrate that the fine-tuned model (Urdu-mT5-mMARCO) achieves a Mean Reciprocal Rank (MRR@10) of 0.247 and a Recall@10 of 0.439, representing significant improvements over zero-shot results and showing the potential for expanding IR access for Urdu speakers. By bridging access gaps for speakers of low-resource languages, this work not only advances multilingual IR research but also emphasizes the ethical and societal importance of inclusive IR technologies. This work provides valuable insights into the challenges and solutions for improving language representation and lays the groundwork for future research, especially in South Asian languages, which can benefit from the adaptable methods used in this study.
Authors: Zihan Liu, Yang Chen, Mohammad Shoeybi, Bryan Catanzaro, Wei Ping
Abstract: In this paper, we introduce AceMath, a suite of frontier math models that excel in solving complex math problems, along with highly effective reward models capable of evaluating generated solutions and reliably identifying the correct ones. To develop the instruction-tuned math models, we propose a supervised fine-tuning (SFT) process that first achieves competitive performance across general domains, followed by targeted fine-tuning for the math domain using a carefully curated set of prompts and synthetically generated responses. The resulting model, AceMath-72B-Instruct greatly outperforms Qwen2.5-Math-72B-Instruct, GPT-4o and Claude-3.5 Sonnet. To develop math-specialized reward model, we first construct AceMath-RewardBench, a comprehensive and robust benchmark for evaluating math reward models across diverse problems and difficulty levels. After that, we present a systematic approach to build our math reward models. The resulting model, AceMath-72B-RM, consistently outperforms state-of-the-art reward models. Furthermore, when combining AceMath-72B-Instruct with AceMath-72B-RM, we achieve the highest average rm@8 score across the math reasoning benchmarks. We release model weights, training data, and evaluation benchmarks at: https://research.nvidia.com/labs/adlr/acemath
Authors: Chris Samarinas, Alexander Krubner, Alireza Salemi, Youngwoo Kim, Hamed Zamani
Abstract: This paper presents ICAT, an evaluation framework for measuring coverage of diverse factual information in long-form text generation. ICAT breaks down a long output text into a list of atomic claims and not only verifies each claim through retrieval from a (reliable) knowledge source, but also computes the alignment between the atomic factual claims and various aspects expected to be presented in the output. We study three implementations of the ICAT framework, each with a different assumption on the availability of aspects and alignment method. By adopting data from the diversification task in the TREC Web Track and the ClueWeb corpus, we evaluate the ICAT framework. We demonstrate strong correlation with human judgments and provide comprehensive evaluation across multiple state-of-the-art LLMs. Our framework further offers interpretable and fine-grained analysis of diversity and coverage. Its modular design allows for easy adaptation to different domains and datasets, making it a valuable tool for evaluating the qualitative aspects of long-form responses produced by LLMs.
Authors: Shreya Havaldar, Hamidreza Alvari, John Palowitch, Mohammad Javad Hosseini, Senaka Buthpitiya, Alex Fabrikant
Abstract: Much of human communication depends on implication, conveying meaning beyond literal words to express a wider range of thoughts, intentions, and feelings. For models to better understand and facilitate human communication, they must be responsive to the text's implicit meaning. We focus on Natural Language Inference (NLI), a core tool for many language tasks, and find that state-of-the-art NLI models and datasets struggle to recognize a range of cases where entailment is implied, rather than explicit from the text. We formalize implied entailment as an extension of the NLI task and introduce the Implied NLI dataset (INLI) to help today's LLMs both recognize a broader variety of implied entailments and to distinguish between implicit and explicit entailment. We show how LLMs fine-tuned on INLI understand implied entailment and can generalize this understanding across datasets and domains.
Authors: Hajime Kiyama, Taichi Aida, Mamoru Komachi, Toshinobu Ogiso, Hiroya Takamura, Daichi Mochihashi
Abstract: The meanings and relationships of words shift over time. This phenomenon is referred to as semantic shift. Research focused on understanding how semantic shifts occur over multiple time periods is essential for gaining a detailed understanding of semantic shifts. However, detecting change points only between adjacent time periods is insufficient for analyzing detailed semantic shifts, and using BERT-based methods to examine word sense proportions incurs a high computational cost. To address those issues, we propose a simple yet intuitive framework for how semantic shifts occur over multiple time periods by leveraging a similarity matrix between the embeddings of the same word through time. We compute a diachronic word similarity matrix using fast and lightweight word embeddings across arbitrary time periods, making it deeper to analyze continuous semantic shifts. Additionally, by clustering the similarity matrices for different words, we can categorize words that exhibit similar behavior of semantic shift in an unsupervised manner.
Authors: Philipp Kuehn, Dilara Nadermahmoodi, Markus Bayer, Christian Reuter
Abstract: Public information contains valuable Cyber Threat Intelligence (CTI) that is used to prevent attacks in the future. Ideally, the learnings from previous attacks help to mitigate all those that follow. While there are standards for sharing this information, much of it is shared in non-standardized news articles or blog posts. It is a time-consuming task to monitor online sources for threats and even then, one can never be sure, to use the right sources. Current research propose extractors of Indicators of Compromise from known sources, while the identification of new sources is rarely considered. This paper proposes a focused crawler focused on the CTI domain based on multi-armed bandit ( MAB) and different crawling strategies. It uses SBERT to identify relevant documents, while dynamically adapt its crawling path. We propose a system called ThreatCrawl, which achieve a harvest rate of over 25% and is able to expand its used seed by over 300%, while retaining focus on the topic at hand. In addition, this crawler identified previously unknown but highly relevant overview pages, datasets, and domains.
Authors: Qiming Bao, Juho Leinonen, Alex Yuxuan Peng, Wanjun Zhong, Ga\"el Gendron, Timothy Pistotti, Alice Huang, Paul Denny, Michael Witbrock, Jiamou Liu
Abstract: Large language models exhibit superior capabilities in processing and understanding language, yet their applications in educational contexts remain underexplored. Learnersourcing enhances learning by engaging students in creating their own educational content. When learnersourcing multiple-choice questions, creating explanations for the solution of a question is a crucial step; it helps other students understand the solution and promotes a deeper understanding of related concepts. However, it is often difficult for students to craft effective solution explanations, due to limited subject understanding. To help scaffold the task of automated explanation generation, we present and evaluate a framework called "ILearner-LLM", that iteratively enhances the generated explanations for the given questions with large language models. Comprising an explanation generation model and an explanation evaluation model, the framework generates high-quality student-aligned explanations by iteratively feeding the quality rating score from the evaluation model back into the instruction prompt of the explanation generation model. Experimental results demonstrate the effectiveness of our ILearner-LLM on LLaMA2-13B and GPT-4 to generate higher quality explanations that are closer to those written by students on five PeerWise datasets. Our findings represent a promising path to enrich the learnersourcing experience for students and to enhance the capabilities of large language models for educational applications.
Authors: Kaifeng Lyu, Haoyu Zhao, Xinran Gu, Dingli Yu, Anirudh Goyal, Sanjeev Arora
Abstract: Public LLMs such as the Llama 2-Chat underwent alignment training and were considered safe. Recently Qi et al. [2024] reported that even benign fine-tuning on seemingly safe datasets can give rise to unsafe behaviors in the models. The current paper is about methods and best practices to mitigate such loss of alignment. We focus on the setting where a public model is fine-tuned before serving users for specific usage, where the model should improve on the downstream task while maintaining alignment. Through extensive experiments on several chat models (Meta's Llama 2-Chat, Mistral AI's Mistral 7B Instruct v0.2, and OpenAI's GPT-3.5 Turbo), this paper uncovers that the prompt templates used during fine-tuning and inference play a crucial role in preserving safety alignment, and proposes the ``Pure Tuning, Safe Testing'' (PTST) strategy -- fine-tune models without a safety prompt, but include it at test time. This seemingly counterintuitive strategy incorporates an intended distribution shift to encourage alignment preservation. Fine-tuning experiments on GSM8K, ChatDoctor, and OpenOrca show that PTST significantly reduces the rise of unsafe behaviors.
Authors: Xinye Tang, Amir H. Abdi, Jeremias Eichelbaum, Mahan Das, Alex Klein, Nihal Irmak Pakis, William Blum, Daniel L Mace, Tanvi Raja, Namrata Padmanabhan, Ye Xing
Abstract: Data is growing rapidly in volume and complexity. Proficiency in database query languages is pivotal for crafting effective queries. As coding assistants become more prevalent, there is significant opportunity to enhance database query languages. The Kusto Query Language (KQL) is a widely used query language for large semi-structured data such as logs, telemetries, and time-series for big data analytics platforms. This paper introduces NL2KQL an innovative framework that uses large language models (LLMs) to convert natural language queries (NLQs) to KQL queries. The proposed NL2KQL framework includes several key components: Schema Refiner which narrows down the schema to its most pertinent elements; the Few-shot Selector which dynamically selects relevant examples from a few-shot dataset; and the Query Refiner which repairs syntactic and semantic errors in KQL queries. Additionally, this study outlines a method for generating large datasets of synthetic NLQ-KQL pairs which are valid within a specific database contexts. To validate NL2KQL's performance, we utilize an array of online (based on query execution) and offline (based on query parsing) metrics. Through ablation studies, the significance of each framework component is examined, and the datasets used for benchmarking are made publicly available. This work is the first of its kind and is compared with available baselines to demonstrate its effectiveness.
Authors: Zhihui Xie, Jiahui Gao, Lei Li, Zhenguo Li, Qi Liu, Lingpeng Kong
Abstract: The widespread adoption of large language models (LLMs) has raised concerns about their safety and reliability, particularly regarding their vulnerability to adversarial attacks. In this paper, we propose a novel perspective that attributes this vulnerability to reward misspecification during the alignment process. This misspecification occurs when the reward function fails to accurately capture the intended behavior, leading to misaligned model outputs. We introduce a metric ReGap to quantify the extent of reward misspecification and demonstrate its effectiveness and robustness in detecting harmful backdoor prompts. Building upon these insights, we present ReMiss, a system for automated red teaming that generates adversarial prompts in a reward-misspecified space. ReMiss achieves state-of-the-art attack success rates on the AdvBench benchmark against various target aligned LLMs while preserving the human readability of the generated prompts. Furthermore, these attacks on open-source models demonstrate high transferability to closed-source models like GPT-4o and out-of-distribution tasks from HarmBench. Detailed analysis highlights the unique advantages of the proposed reward misspecification objective compared to previous methods, offering new insights for improving LLM safety and robustness.
Authors: Zhisheng Tang, Mayank Kejriwal
Abstract: Spatial reasoning, an important faculty of human cognition with many practical applications, is one of the core commonsense skills that is not purely language-based and, for satisfying (as opposed to optimal) solutions, requires some minimum degree of planning. Existing benchmarks of Commonsense Spatial Reasoning (CSR) tend to evaluate how Large Language Models (LLMs) interpret text-based spatial $\textit{descriptions}$ rather than directly evaluate a plan produced by the LLM in response to a $\textit{specific}$ spatial reasoning problem. In this paper, we construct a large-scale benchmark called GRASP, which consists of 16,000 grid-based environments where the agent is tasked with an energy collection problem. These environments include 100 grid instances instantiated using each of the 160 different grid settings, involving five different energy distributions, two modes of agent starting position, and two distinct obstacle configurations, as well as three kinds of agent constraints. Using GRASP, we compare classic baseline approaches, such as random walk and greedy search methods, with advanced LLMs like GPT-3.5-Turbo, GPT-4o, and GPT-o1-mini. The experimental results indicate that even these advanced LLMs struggle to consistently achieve satisfactory solutions.
Authors: Han Guo, William Brandon, Radostin Cholakov, Jonathan Ragan-Kelley, Eric P. Xing, Yoon Kim
Abstract: The deployment of large language models (LLMs) is often constrained by memory bandwidth, where the primary bottleneck is the cost of transferring model parameters from the GPU's global memory to its registers. When coupled with custom kernels that fuse the dequantization and matmul operations, weight-only quantization can thus enable faster inference by reducing the amount of memory movement. However, developing high-performance kernels for weight-quantized LLMs presents substantial challenges, especially when the weights are compressed to non-evenly-divisible bit widths (e.g., 3 bits) with non-uniform, lookup table (LUT) quantization. This paper describes FLUTE, a flexible lookup table engine for LUT-quantized LLMs, which uses offline restructuring of the quantized weight matrix to minimize bit manipulations associated with unpacking, and vectorization and duplication of the lookup table to mitigate shared memory bandwidth constraints. At batch sizes < 32 and quantization group size of 128 (typical in LLM inference), the FLUTE kernel can be 2-4x faster than existing GEMM kernels. As an application of FLUTE, we explore a simple extension to lookup table-based NormalFloat quantization and apply it to quantize LLaMA3 to various configurations, obtaining competitive quantization performance against strong baselines while obtaining an end-to-end throughput increase of 1.5 to 2 times.
Authors: Kristian Gonz\'alez Barman, Simon Lohse, Henk de Regt
Abstract: We argue for the epistemic and ethical advantages of pluralism in Reinforcement Learning from Human Feedback (RLHF) in the context of Large Language Models (LLM). Drawing on social epistemology and pluralist philosophy of science, we suggest ways in which RHLF can be made more responsive to human needs and how we can address challenges along the way. The paper concludes with an agenda for change, i.e. concrete, actionable steps to improve LLM development.
Authors: Teresa Dorszewski, Albert Kj{\o}ller Jacobsen, Lenka T\v{e}tkov\'a, Lars Kai Hansen
Abstract: Self-supervised speech representation models, particularly those leveraging transformer architectures, have demonstrated remarkable performance across various tasks such as speech recognition, speaker identification, and emotion detection. Recent studies on transformer models revealed a high redundancy between layers and the potential for significant pruning, which we will investigate here for transformer-based speech representation models. We perform a detailed analysis of layer similarity in speech representation models using three similarity metrics: cosine similarity, centered kernel alignment, and mutual nearest-neighbor alignment. Our findings reveal a block-like structure of high similarity, suggesting two main processing steps and significant redundancy of layers. We demonstrate the effectiveness of pruning transformer-based speech representation models without the need for post-training, achieving up to 40% reduction in transformer layers while maintaining over 95% of the model's predictive capacity. Furthermore, we employ a knowledge distillation method to substitute the entire transformer stack with mimicking layers, reducing the network size 95-98% and the inference time by up to 94%. This substantial decrease in computational load occurs without considerable performance loss, suggesting that the transformer stack is almost completely redundant for downstream applications of speech representation models.
Authors: Ziyao Zhang, Yanlin Wang, Chong Wang, Jiachi Chen, Zibin Zheng
Abstract: Code generation aims to automatically generate code from input requirements, significantly enhancing development efficiency. Recent large language models (LLMs) based approaches have shown promising results and revolutionized code generation task. Despite the promising performance, LLMs often generate contents with hallucinations, especially for the code generation scenario requiring the handling of complex contextual dependencies in practical development process. Although previous study has analyzed hallucinations in LLM-powered code generation, the study is limited to standalone function generation. In this paper, we conduct an empirical study to study the phenomena, mechanism, and mitigation of LLM hallucinations within more practical and complex development contexts in repository-level generation scenario. First, we manually examine the code generation results from six mainstream LLMs to establish a hallucination taxonomy of LLM-generated code. Next, we elaborate on the phenomenon of hallucinations, analyze their distribution across different models. We then analyze causes of hallucinations and identify four potential factors contributing to hallucinations. Finally, we propose an RAG-based mitigation method, which demonstrates consistent effectiveness in all studied LLMs. The replication package including code, data, and experimental results is available at https://github.com/DeepSoftwareAnalytics/LLMCodingHallucination
URLs: https://github.com/DeepSoftwareAnalytics/LLMCodingHallucination
Authors: Chen Cai, Zheng Wang, Jianjun Gao, Wenyang Liu, Ye Lu, Runzhong Zhang, Kim-Hui Yap
Abstract: In recent years, the rapid increase in online video content has underscored the limitations of static Video Question Answering (VideoQA) models trained on fixed datasets, as they struggle to adapt to new questions or tasks posed by newly available content. In this paper, we explore the novel challenge of VideoQA within a continual learning framework, and empirically identify a critical issue: fine-tuning a large language model (LLM) for a sequence of tasks often results in catastrophic forgetting. To address this, we propose Collaborative Prompting (ColPro), which integrates specific question constraint prompting, knowledge acquisition prompting, and visual temporal awareness prompting. These prompts aim to capture textual question context, visual content, and video temporal dynamics in VideoQA, a perspective underexplored in prior research. Experimental results on the NExT-QA and DramaQA datasets show that ColPro achieves superior performance compared to existing approaches, achieving 55.14\% accuracy on NExT-QA and 71.24\% accuracy on DramaQA, highlighting its practical relevance and effectiveness.
Authors: Or Ordentlich, Yury Polyanskiy
Abstract: Recent work in machine learning community proposed multiple methods for performing lossy compression (quantization) of large matrices. This quantization is important for accelerating matrix multiplication (main component of large language models), which is often bottlenecked by the speed of loading these matrices from memory. Unlike classical vector quantization and rate-distortion theory, the goal of these new compression algorithms is to be able to approximate not the matrices themselves, but their matrix product. Specifically, given a pair of real matrices $A,B$ an encoder (compressor) is applied to each of them independently producing descriptions with $R$ bits per entry. These representations subsequently are used by the decoder to estimate matrix product $A^\top B$. In this work, we provide a non-asymptotic lower bound on the mean squared error of this approximation (as a function of rate $R$) for the case of matrices $A,B$ with iid Gaussian entries. Algorithmically, we construct a universal quantizer based on nested lattices with an explicit guarantee of approximation error for any (non-random) pair of matrices $A$, $B$ in terms of only Frobenius norms $\|\bar{A}\|_F, \|\bar{B}\|_F$ and $\|\bar{A}^\top \bar{B}\|_F$, where $\bar{A},\bar{B}$ are versions of $A,B$ with zero-centered columns, respectively. For iid Gaussian matrices our quantizer achieves the lower bound and is, thus, asymptotically optimal. A practical low-complexity version of our quantizer achieves performance quite close to optimal. In addition, we derive rate-distortion function for matrix multiplication of iid Gaussian matrices, which exhibits an interesting phase-transition at $R\approx 0.906$ bit/entry.
Authors: Yukun Cao, Lisheng Wang, Luobin Huang
Abstract: Temporal knowledge graph (TKG) reasoning that infers future missing facts is an essential and challenging task. Predicting future events typically relies on closely related historical facts, yielding more accurate results for repetitive or periodic events. However, for future events with sparse historical interactions, the effectiveness of this method, which focuses on leveraging high-frequency historical information, diminishes. Recently, the capabilities of diffusion models in image generation have opened new opportunities for TKG reasoning. Therefore, we propose a graph node diffusion model with dual-domain periodic contrastive learning (DPCL-Diff). Graph node diffusion model (GNDiff) introduces noise into sparsely related events to simulate new events, generating high-quality data that better conforms to the actual distribution. This generative mechanism significantly enhances the model's ability to reason about new events. Additionally, the dual-domain periodic contrastive learning (DPCL) maps periodic and non-periodic event entities to Poincar\'e and Euclidean spaces, leveraging their characteristics to distinguish similar periodic events effectively. Experimental results on four public datasets demonstrate that DPCL-Diff significantly outperforms state-of-the-art TKG models in event prediction, demonstrating our approach's effectiveness. This study also investigates the combined effectiveness of GNDiff and DPCL in TKG tasks.
Authors: Yukun Cao, Zengyi Gao, Zhiyang Li, Xike Xie, Kevin Zhou, Jianliang Xu
Abstract: GraphRAG integrates (knowledge) graphs with large language models (LLMs) to improve reasoning accuracy and contextual relevance. Despite its promising applications and strong relevance to multiple research communities, such as databases and natural language processing, GraphRAG currently lacks modular workflow analysis, systematic solution frameworks, and insightful empirical studies. To bridge these gaps, we propose LEGO-GraphRAG, a modular framework that enables: 1) fine-grained decomposition of the GraphRAG workflow, 2) systematic classification of existing techniques and implemented GraphRAG instances, and 3) creation of new GraphRAG instances. Our framework facilitates comprehensive empirical studies of GraphRAG on large-scale real-world graphs and diverse query sets, revealing insights into balancing reasoning quality, runtime efficiency, and token or GPU cost, that are essential for building advanced GraphRAG systems.
Authors: Xuhao Hu, Dongrui Liu, Hao Li, Xuanjing Huang, Jing Shao
Abstract: Safety concerns of Multimodal large language models (MLLMs) have gradually become an important problem in various applications. Surprisingly, previous works indicate a counter-intuitive phenomenon that using textual unlearning to align MLLMs achieves comparable safety performances with MLLMs trained with image-text pairs. To explain such a counter-intuitive phenomenon, we discover a visual safety information leakage (VSIL) problem in existing multimodal safety benchmarks, i.e., the potentially risky and sensitive content in the image has been revealed in the textual query. In this way, MLLMs can easily refuse these sensitive text-image queries according to textual queries. However, image-text pairs without VSIL are common in real-world scenarios and are overlooked by existing multimodal safety benchmarks. To this end, we construct multimodal visual leakless safety benchmark (VLSBench) preventing visual safety leakage from image to textual query with 2.4k image-text pairs. Experimental results indicate that VLSBench poses a significant challenge to both open-source and close-source MLLMs, including LLaVA, Qwen2-VL, Llama3.2-Vision, and GPT-4o. This study demonstrates that textual alignment is enough for multimodal safety scenarios with VSIL, while multimodal alignment is a more promising solution for multimodal safety scenarios without VSIL. Please see our code and data at: https://hxhcreate.github.io/vlsbench.github.io/
Authors: Jiliang Hu, Zuchao Li, Mengjia Shen, Haojun Ai, Sheng Li, Jun Zhang
Abstract: Spoken language understanding (SLU) is a structure prediction task in the field of speech. Recently, many works on SLU that treat it as a sequence-to-sequence task have achieved great success. However, This method is not suitable for simultaneous speech recognition and understanding. In this paper, we propose a joint speech recognition and structure learning framework (JSRSL), an end-to-end SLU model based on span, which can accurately transcribe speech and extract structured content simultaneously. We conduct experiments on name entity recognition and intent classification using the Chinese dataset AISHELL-NER and the English dataset SLURP. The results show that our proposed method not only outperforms the traditional sequence-to-sequence method in both transcription and extraction capabilities but also achieves state-of-the-art performance on the two datasets.
Authors: Fengli Xu, Qianyue Hao, Zefang Zong, Jingwei Wang, Yunke Zhang, Jingyi Wang, Xiaochong Lan, Jiahui Gong, Tianjian Ouyang, Fanjin Meng, Chenyang Shao, Yuwei Yan, Qinglong Yang, Yiwen Song, Sijian Ren, Xinyuan Hu, Yu Li, Jie Feng, Chen Gao, Yong Li
Abstract: Language has long been conceived as an essential tool for human reasoning. The breakthrough of Large Language Models (LLMs) has sparked significant research interest in leveraging these models to tackle complex reasoning tasks. Researchers have moved beyond simple autoregressive token generation by introducing the concept of "thought" -- a sequence of tokens representing intermediate steps in the reasoning process. This innovative paradigm enables LLMs' to mimic complex human reasoning processes, such as tree search and reflective thinking. Recently, an emerging trend of learning to reason has applied reinforcement learning (RL) to train LLMs to master reasoning processes. This approach enables the automatic generation of high-quality reasoning trajectories through trial-and-error search algorithms, significantly expanding LLMs' reasoning capacity by providing substantially more training data. Furthermore, recent studies demonstrate that encouraging LLMs to "think" with more tokens during test-time inference can further significantly boost reasoning accuracy. Therefore, the train-time and test-time scaling combined to show a new research frontier -- a path toward Large Reasoning Model. The introduction of OpenAI's o1 series marks a significant milestone in this research direction. In this survey, we present a comprehensive review of recent progress in LLM reasoning. We begin by introducing the foundational background of LLMs and then explore the key technical components driving the development of large reasoning models, with a focus on automated data construction, learning-to-reason techniques, and test-time scaling. We also analyze popular open-source projects at building large reasoning models, and conclude with open challenges and future research directions.