Authors: Jo\~ao Pedro Gandarela, Danilo S. Carvalho, Andr\'e Freitas
Abstract: This work presents a novel systematic methodology to analyse the capabilities and limitations of Large Language Models (LLMs) with feedback from a formal inference engine, on logic theory induction. The analysis is complexity-graded w.r.t. rule dependency structure, allowing quantification of specific inference challenges on LLM performance. Integrating LLMs with formal methods is a promising frontier in the Natural Language Processing field, as an important avenue for improving model inference control and explainability. In particular, inductive learning over complex sets of facts and rules, poses unique challenges for current autoregressive models, as they lack explicit symbolic grounding. While they can be complemented by formal systems, the properties delivered by LLMs regarding inductive learning, are not well understood and quantified. Empirical results indicate that the largest LLMs can achieve competitive results against a SOTA Inductive Logic Programming (ILP) system baseline, but also that tracking long predicate relationship chains is a more difficult obstacle than theory complexity for the LLMs.
Authors: Kasper Cools, Gideon Mailette de Buy Wenniger, Clara Maathuis
Abstract: The advent of social media transformed interpersonal communication and information consumption processes. This digital landscape accommodates user intentions, also resulting in an increase of offensive language and harmful behavior. Concurrently, social media platforms collect vast datasets comprising user-generated content and behavioral information. These datasets are instrumental for platforms deploying machine learning and data-driven strategies, facilitating customer insights and countermeasures against social manipulation mechanisms like disinformation and offensive content. Nevertheless, the availability of such datasets, along with the application of various machine learning techniques, to researchers and practitioners, for specific social media platforms regarding particular events, is limited. In particular for TikTok, which offers unique tools for personalized content creation and sharing, the existing body of knowledge would benefit from having diverse comprehensive datasets and associated data analytics solutions on offensive content. While efforts from social media platforms, research, and practitioner communities are seen on this behalf, such content continues to proliferate. This translates to an essential need to make datasets publicly available and build corresponding intelligent solutions. On this behalf, this research undertakes the collection and analysis of TikTok data containing offensive content, building a series of machine learning and deep learning models for offensive content detection. This is done aiming at answering the following research question: "How to develop a series of computational models to detect offensive content on TikTok?". To this end, a Data Science methodological approach is considered, 120.423 TikTok comments are collected, and on a balanced, binary classification approach, F1 score performance results of 0.863 is obtained.
Authors: Fnu Mohbat, Mohammed J. Zaki
Abstract: In the rapidly evolving landscape of online recipe sharing within a globalized context, there has been a notable surge in research towards comprehending and generating food recipes. Recent advancements in large language models (LLMs) like GPT-2 and LLaVA have paved the way for Natural Language Processing (NLP) approaches to delve deeper into various facets of food-related tasks, encompassing ingredient recognition and comprehensive recipe generation. Despite impressive performance and multi-modal adaptability of LLMs, domain-specific training remains paramount for their effective application. This work evaluates existing LLMs for recipe generation and proposes LLaVA-Chef, a novel model trained on a curated dataset of diverse recipe prompts in a multi-stage approach. First, we refine the mapping of visual food image embeddings to the language space. Second, we adapt LLaVA to the food domain by fine-tuning it on relevant recipe data. Third, we utilize diverse prompts to enhance the model's recipe comprehension. Finally, we improve the linguistic quality of generated recipes by penalizing the model with a custom loss function. LLaVA-Chef demonstrates impressive improvements over pretrained LLMs and prior works. A detailed qualitative analysis reveals that LLaVA-Chef generates more detailed recipes with precise ingredient mentions, compared to existing approaches.
Authors: Ond\v{r}ej Pra\v{z}\'ak, Miloslav Konop\'ik
Abstract: Coreference resolution, the task of identifying expressions in text that refer to the same entity, is a critical component in various natural language processing (NLP) applications. This paper presents our end-to-end neural coreference resolution system, utilizing the CorefUD 1.1 dataset, which spans 17 datasets across 12 languages. We first establish strong baseline models, including monolingual and cross-lingual variations, and then propose several extensions to enhance performance across diverse linguistic contexts. These extensions include cross-lingual training, incorporation of syntactic information, a Span2Head model for optimized headword prediction, and advanced singleton modeling. We also experiment with headword span representation and long-documents modeling through overlapping segments. The proposed extensions, particularly the heads-only approach, singleton modeling, and long document prediction significantly improve performance across most datasets. We also perform zero-shot cross-lingual experiments, highlighting the potential and limitations of cross-lingual transfer in coreference resolution. Our findings contribute to the development of robust and scalable coreference systems for multilingual coreference resolution. Finally, we evaluate our model on CorefUD 1.1 test set and surpass the best model from CRAC 2023 shared task of a comparable size by a large margin. Our nodel is available on GitHub: \url{https://github.com/ondfa/coref-multiling}
Authors: Lu\'is Filipe Cunha, Purifica\c{c}\~ao Silvano, Ricardo Campos, Al\'ipio Jorge
Abstract: Event extraction is an NLP task that commonly involves identifying the central word (trigger) for an event and its associated arguments in text. ACE-2005 is widely recognised as the standard corpus in this field. While other corpora, like PropBank, primarily focus on annotating predicate-argument structure, ACE-2005 provides comprehensive information about the overall event structure and semantics. However, its limited language coverage restricts its usability. This paper introduces ACE-2005-PT, a corpus created by translating ACE-2005 into Portuguese, with European and Brazilian variants. To speed up the process of obtaining ACE-2005-PT, we rely on automatic translators. This, however, poses some challenges related to automatically identifying the correct alignments between multi-word annotations in the original text and in the corresponding translated sentence. To achieve this, we developed an alignment pipeline that incorporates several alignment techniques: lemmatization, fuzzy matching, synonym matching, multiple translations and a BERT-based word aligner. To measure the alignment effectiveness, a subset of annotations from the ACE-2005-PT corpus was manually aligned by a linguist expert. This subset was then compared against our pipeline results which achieved exact and relaxed match scores of 70.55\% and 87.55\% respectively. As a result, we successfully generated a Portuguese version of the ACE-2005 corpus, which has been accepted for publication by LDC.
Authors: Lu\'is Filipe Cunha, Ricardo Campos, Al\'ipio Jorge
Abstract: Event extraction is an Information Retrieval task that commonly consists of identifying the central word for the event (trigger) and the event's arguments. This task has been extensively studied for English but lags behind for Portuguese, partly due to the lack of task-specific annotated corpora. This paper proposes a framework in which two separated BERT-based models were fine-tuned to identify and classify events in Portuguese documents. We decompose this task into two sub-tasks. Firstly, we use a token classification model to detect event triggers. To extract event arguments, we train a Question Answering model that queries the triggers about their corresponding event argument roles. Given the lack of event annotated corpora in Portuguese, we translated the original version of the ACE-2005 dataset (a reference in the field) into Portuguese, producing a new corpus for Portuguese event extraction. To accomplish this, we developed an automatic translation pipeline. Our framework obtains F1 marks of 64.4 for trigger classification and 46.7 for argument classification setting, thus a new state-of-the-art reference for these tasks in Portuguese.
Authors: Chong Shen, Chenyue Zhou
Abstract: In this work, we investigate the effectiveness of injecting external knowledge to a large language model (LLM) to identify semantic plausibility of simple events. Specifically, we enhance the LLM with fine-grained entity types, event types and their definitions extracted from an external knowledge base. These knowledge are injected into our system via designed templates. We also augment the data to balance the label distribution and adapt the task setting to real world scenarios in which event mentions are expressed as natural language sentences. The experimental results show the effectiveness of the injected knowledge on modeling semantic plausibility of events. An error analysis further emphasizes the importance of identifying non-trivial entity and event types.
Authors: Chen Wang, Rohitash Chandra
Abstract: The COVID-19 pandemic has exacerbated xenophobia, particularly Sinophobia, leading to widespread discrimination against individuals of Chinese descent. Large language models (LLMs) are pre-trained deep learning models used for natural language processing (NLP) tasks. The ability of LLMs to understand and generate human-like text makes them particularly useful for analysing social media data to detect and evaluate sentiments. We present a sentiment analysis framework utilising LLMs for longitudinal sentiment analysis of the Sinophobic sentiments expressed in X (Twitter) during the COVID-19 pandemic. The results show a significant correlation between the spikes in Sinophobic tweets, Sinophobic sentiments and surges in COVID-19 cases, revealing that the evolution of the pandemic influenced public sentiment and the prevalence of Sinophobic discourse. Furthermore, the sentiment analysis revealed a predominant presence of negative sentiments, such as annoyance and denial, which underscores the impact of political narratives and misinformation shaping public opinion. The lack of empathetic sentiment which was present in previous studies related to COVID-19 highlights the way the political narratives in media viewed the pandemic and how it blamed the Chinese community. Our study highlights the importance of transparent communication in mitigating xenophobic sentiments during global crises.
Authors: Weijie Liu, Zecheng Tang, Juntao Li, Kehai Chen, Min Zhang
Abstract: Recent advancements in Large Language Models (LLMs) have yielded remarkable success across diverse fields. However, handling long contexts remains a significant challenge for LLMs due to the quadratic time and space complexity of attention mechanisms and the growing memory consumption of the key-value cache during generation. This work introduces MemLong: Memory-Augmented Retrieval for Long Text Generation, a method designed to enhance the capabilities of long-context language modeling by utilizing an external retriever for historical information retrieval. MemLong combines a non-differentiable ``ret-mem'' module with a partially trainable decoder-only language model and introduces a fine-grained, controllable retrieval attention mechanism that leverages semantic-level relevant chunks. Comprehensive evaluations on multiple long-context language modeling benchmarks demonstrate that MemLong consistently outperforms other state-of-the-art LLMs. More importantly, MemLong can extend the context length on a single 3090 GPU from 4k up to 80k. Our code is available at https://github.com/Bui1dMySea/MemLong
Authors: Zhongyuan Wang, Richong Zhang, Zhijie Nie, Jaein Kim
Abstract: Recent Text-to-SQL methods leverage large language models (LLMs) by incorporating feedback from the database management system. While these methods effectively address execution errors in SQL queries, they struggle with database mismatches -- errors that do not trigger execution exceptions. Database mismatches include issues such as condition mismatches and stricter constraint mismatches, both of which are more prevalent in real-world scenarios. To address these challenges, we propose a tool-assisted agent framework for SQL inspection and refinement, equipping the LLM-based agent with two specialized tools: a retriever and a detector, designed to diagnose and correct SQL queries with database mismatches. These tools enhance the capability of LLMs to handle real-world queries more effectively. We also introduce Spider-Mismatch, a new dataset specifically constructed to reflect the condition mismatch problems encountered in real-world scenarios. Experimental results demonstrate that our method achieves the highest performance on the averaged results of the Spider and Spider-Realistic datasets in few-shot settings, and it significantly outperforms baseline methods on the more realistic dataset, Spider-Mismatch.
Authors: Guangya Wan, Yuqi Wu, Jie Chen, Sheng Li
Abstract: Self-Consistency (SC) is a widely used method to mitigate hallucinations in Large Language Models (LLMs) by sampling the LLM multiple times and outputting the most frequent solution. Despite its benefits, SC results in significant computational costs proportional to the number of samples generated. Previous early-stopping approaches, such as Early Stopping Self Consistency and Adaptive Consistency, have aimed to reduce these costs by considering output consistency, but they do not analyze the quality of the reasoning paths (RPs) themselves. To address this issue, we propose Reasoning-Aware Self-Consistency (RASC), an innovative early-stopping framework that dynamically adjusts the number of sample generations by considering both the output answer and the RPs from Chain of Thought (CoT) prompting. RASC assigns confidence scores sequentially to the generated samples, stops when certain criteria are met, and then employs weighted majority voting to optimize sample usage and enhance answer reliability. We comprehensively test RASC with multiple LLMs across varied QA datasets. RASC outperformed existing methods and significantly reduces sample usage by an average of 80% while maintaining or improving accuracy up to 5% compared to the original SC
Authors: Atnafu Lambebo Tonja, Bonaventure F. P. Dossou, Jessica Ojo, Jenalea Rajab, Fadel Thior, Eric Peter Wairagala, Aremu Anuoluwapo, Pelonomi Moiloa, Jade Abbott, Vukosi Marivate, Benjamin Rosman
Abstract: High-resource language models often fall short in the African context, where there is a critical need for models that are efficient, accessible, and locally relevant, even amidst significant computing and data constraints. This paper introduces InkubaLM, a small language model with 0.4 billion parameters, which achieves performance comparable to models with significantly larger parameter counts and more extensive training data on tasks such as machine translation, question-answering, AfriMMLU, and the AfriXnli task. Notably, InkubaLM outperforms many larger models in sentiment analysis and demonstrates remarkable consistency across multiple languages. This work represents a pivotal advancement in challenging the conventional paradigm that effective language models must rely on substantial resources. Our model and datasets are publicly available \footnote{\url{https://huggingface.co/lelapa}} to encourage research and development on low-resource languages.
Authors: Minxue Niu (University of Michigan), Mimansa Jaiswal (Independent Researcher), Emily Mower Provost (University of Michigan)
Abstract: Training emotion recognition models has relied heavily on human annotated data, which present diversity, quality, and cost challenges. In this paper, we explore the potential of Large Language Models (LLMs), specifically GPT4, in automating or assisting emotion annotation. We compare GPT4 with supervised models and or humans in three aspects: agreement with human annotations, alignment with human perception, and impact on model training. We find that common metrics that use aggregated human annotations as ground truth can underestimate the performance, of GPT-4 and our human evaluation experiment reveals a consistent preference for GPT-4 annotations over humans across multiple datasets and evaluators. Further, we investigate the impact of using GPT-4 as an annotation filtering process to improve model training. Together, our findings highlight the great potential of LLMs in emotion annotation tasks and underscore the need for refined evaluation methodologies.
Authors: Maxime M\'eloux, Christophe Cerisara
Abstract: Teaching new information to pre-trained large language models (PLM) is a crucial but challenging task. Model adaptation techniques, such as fine-tuning and parameter-efficient training have been shown to store new facts at a slow rate; continual learning is an option but is costly and prone to catastrophic forgetting. This work studies and quantifies how PLM may learn and remember new world knowledge facts that do not occur in their pre-training corpus, which only contains world knowledge up to a certain date. To that purpose, we first propose Novel-WD, a new dataset consisting of sentences containing novel facts extracted from recent Wikidata updates, along with two evaluation tasks in the form of causal language modeling and multiple choice questions (MCQ). We make this dataset freely available to the community, and release a procedure to later build new versions of similar datasets with up-to-date information. We also explore the use of prefix-tuning for novel information learning, and analyze how much information can be stored within a given prefix. We show that a single fact can reliably be encoded within a single prefix, and that the prefix capacity increases with its length and with the base model size.
Authors: Yujing Wang, Hainan Zhang, Liang Pang, Liang Pang, Hongwei Zheng, Zhiming Zheng
Abstract: In a real-world RAG system, the current query often involves spoken ellipses and ambiguous references from dialogue contexts, necessitating query rewriting to better describe user's information needs. However, traditional context-based rewriting has minimal enhancement on downstream generation tasks due to the lengthy process from query rewriting to response generation. Some researchers try to utilize reinforcement learning with generation feedback to assist the rewriter, but these sparse rewards provide little guidance in most cases, leading to unstable training and generation results. We find that user's needs are also reflected in the gold document, retrieved documents and ground truth. Therefore, by feeding back these multi-aspect dense rewards to query rewriting, more stable and satisfactory responses can be achieved. In this paper, we propose a novel query rewriting method MaFeRw, which improves RAG performance by integrating multi-aspect feedback from both the retrieval process and generated results. Specifically, we first use manual data to train a T5 model for the rewriter initialization. Next, we design three metrics as reinforcement learning feedback: the similarity between the rewritten query and the gold document, the ranking metrics, and ROUGE between the generation and the ground truth. Inspired by RLAIF, we train three kinds of reward models for the above metrics to achieve more efficient training. Finally, we combine the scores of these reward models as feedback, and use PPO algorithm to explore the optimal query rewriting strategy. Experimental results on two conversational RAG datasets demonstrate that MaFeRw achieves superior generation metrics and more stable training compared to baselines.
Authors: Shubham Agarwal, Thomas Searle, Mart Ratas, Anthony Shek, James Teo, Richard Dobson
Abstract: Electronic Health Records are large repositories of valuable clinical data, with a significant portion stored in unstructured text format. This textual data includes clinical events (e.g., disorders, symptoms, findings, medications and procedures) in context that if extracted accurately at scale can unlock valuable downstream applications such as disease prediction. Using an existing Named Entity Recognition and Linking methodology, MedCAT, these identified concepts need to be further classified (contextualised) for their relevance to the patient, and their temporal and negated status for example, to be useful downstream. This study performs a comparative analysis of various natural language models for medical text classification. Extensive experimentation reveals the effectiveness of transformer-based language models, particularly BERT. When combined with class imbalance mitigation techniques, BERT outperforms Bi-LSTM models by up to 28% and the baseline BERT model by up to 16% for recall of the minority classes. The method has been implemented as part of CogStack/MedCAT framework and made available to the community for further research.
Authors: Esther Ploeger, Huiyuan Lai, Rik van Noord, Antonio Toral
Abstract: Machine translations are found to be lexically poorer than human translations. The loss of lexical diversity through MT poses an issue in the automatic translation of literature, where it matters not only what is written, but also how it is written. Current methods for increasing lexical diversity in MT are rigid. Yet, as we demonstrate, the degree of lexical diversity can vary considerably across different novels. Thus, rather than aiming for the rigid increase of lexical diversity, we reframe the task as recovering what is lost in the machine translation process. We propose a novel approach that consists of reranking translation candidates with a classifier that distinguishes between original and translated text. We evaluate our approach on 31 English-to-Dutch book translations, and find that, for certain books, our approach retrieves lexical diversity scores that are close to human translation.
Authors: Shaojun Xu, Xiaohui Ye, Mengqi Zhang, Pei Wang
Abstract: We apply a state-of-the-art difference-in-differences approach to estimate the impact of ChatGPT's release on the writing style of condensed matter papers on arXiv. Our analysis reveals a statistically significant improvement in the English quality of abstracts written by non-native English speakers. Importantly, this improvement remains robust even after accounting for other potential factors, confirming that it can be attributed to the release of ChatGPT. This indicates widespread adoption of the tool. Following the release of ChatGPT, there is a significant increase in the use of unique words, while the frequency of rare words decreases. Across language families, the changes in writing style are significant for authors from the Latin and Ural-Altaic groups, but not for those from the Germanic or other Indo-European groups.
Authors: Francesca Grasso, Stefano Locci
Abstract: This paper examines the performance of two Large Language Models (LLMs), GPT3.5 and Llama2 and one Small Language Model (SLM) Gemma, across three different classification tasks within the climate change (CC) and environmental domain. Employing BERT-based models as a baseline, we compare their efficacy against these transformer-based models. Additionally, we assess the models' self-evaluation capabilities by analyzing the calibration of verbalized confidence scores in these text classification tasks. Our findings reveal that while BERT-based models generally outperform both the LLMs and SLM, the performance of the large generative models is still noteworthy. Furthermore, our calibration analysis reveals that although Gemma is well-calibrated in initial tasks, it thereafter produces inconsistent results; Llama is reasonably calibrated, and GPT consistently exhibits strong calibration. Through this research, we aim to contribute to the ongoing discussion on the utility and effectiveness of generative LMs in addressing some of the planet's most urgent issues, highlighting their strengths and limitations in the context of ecology and CC.
Authors: Junhao Ruan, Abudukeyumu Abudula, Xinyu Liu, Bei Li, Yinqiao Li, Chenglong Wang, Yuchun Fan, Yuan Ge, Tong Xiao, Jingbo Zhu
Abstract: Large language models (LLMs) trained on next-token prediction (NTP) paradigm have demonstrated powerful capabilities. However, the existing NTP paradigm contains several limitations, particularly related to planned task complications and error propagation during inference. In our work, we extend the critique of NTP, highlighting its limitation also due to training with a narrow objective: the prediction of a sub-optimal one-hot distribution. To support this critique, we conducted a pre-experiment treating the output distribution from powerful LLMs as efficient world data compression. By evaluating the similarity between the $n$-gram distribution and the one-hot distribution with LLMs, we observed that the $n$-gram distributions align more closely with the output distribution of LLMs. Based on this insight, we introduce Next Distribution Prediction (NDP), which uses $n$-gram distributions to replace the one-hot targets, enhancing learning without extra online training time. We conducted experiments across translation, general task, language transfer, and medical domain adaptation. Compared to NTP, NDP can achieve up to +2.97 COMET improvement in translation tasks, +0.61 average improvement in general tasks, and incredible +10.75 average improvement in the medical domain. This demonstrates the concrete benefits of addressing the target narrowing problem, pointing to a new direction for future work on improving NTP.
Authors: Jonathan Bourne
Abstract: The digitisation of historical print media archives is crucial for increasing accessibility to contemporary records. However, the process of Optical Character Recognition (OCR) used to convert physical records to digital text is prone to errors, particularly in the case of newspapers and periodicals due to their complex layouts. This paper introduces Context Leveraging OCR Correction (CLOCR-C), which utilises the infilling and context-adaptive abilities of transformer-based language models (LMs) to improve OCR quality. The study aims to determine if LMs can perform post-OCR correction, improve downstream NLP tasks, and the value of providing the socio-cultural context as part of the correction process. Experiments were conducted using seven LMs on three datasets: the 19th Century Serials Edition (NCSE) and two datasets from the Overproof collection. The results demonstrate that some LMs can significantly reduce error rates, with the top-performing model achieving over a 60% reduction in character error rate on the NCSE dataset. The OCR improvements extend to downstream tasks, such as Named Entity Recognition, with increased Cosine Named Entity Similarity. Furthermore, the study shows that providing socio-cultural context in the prompts improves performance, while misleading prompts lower performance. In addition to the findings, this study releases a dataset of 91 transcribed articles from the NCSE, containing a total of 40 thousand words, to support further research in this area. The findings suggest that CLOCR-C is a promising approach for enhancing the quality of existing digital archives by leveraging the socio-cultural information embedded in the LMs and the text requiring correction.
Authors: Raoyuan Zhao, Abdullatif K\"oksal, Yihong Liu, Leonie Weissweiler, Anna Korhonen, Hinrich Sch\"utze
Abstract: Traditional benchmarking in NLP typically involves using static held-out test sets. However, this approach often results in an overestimation of performance and lacks the ability to offer comprehensive, interpretable, and dynamic assessments of NLP models. Recently, works like DynaBench (Kiela et al., 2021) and CheckList (Ribeiro et al., 2020) have addressed these limitations through behavioral testing of NLP models with test types generated by a multistep human-annotated pipeline. Unfortunately, manually creating a variety of test types requires much human labor, often at prohibitive cost. In this work, we propose SYNTHEVAL, a hybrid behavioral testing framework that leverages large language models (LLMs) to generate a wide range of test types for a comprehensive evaluation of NLP models. SYNTHEVAL first generates sentences via LLMs using controlled generation, and then identifies challenging examples by comparing the predictions made by LLMs with task-specific NLP models. In the last stage, human experts investigate the challenging examples, manually design templates, and identify the types of failures the taskspecific models consistently exhibit. We apply SYNTHEVAL to two classification tasks, sentiment analysis and toxic language detection, and show that our framework is effective in identifying weaknesses of strong models on these tasks. We share our code in https://github.com/Loreley99/SynthEval_CheckList.
Authors: Qian Cao, Xu Chen, Ruihua Song, Xiting Wang, Xinting Huang, Yuchen Ren
Abstract: Image captioning, which generates natural language descriptions of the visual information in an image, is a crucial task in vision-language research. Previous models have typically addressed this task by aligning the generative capabilities of machines with human intelligence through statistical fitting of existing datasets. While effective for normal images, they may struggle to accurately describe those where certain parts of the image are obscured or edited, unlike humans who excel in such cases. These weaknesses they exhibit, including hallucinations and limited interpretability, often hinder performance in scenarios with shifted association patterns. In this paper, we present a generic image captioning framework that employs causal inference to make existing models more capable of interventional tasks, and counterfactually explainable. Our approach includes two variants leveraging either total effect or natural direct effect. Integrating them into the training process enables models to handle counterfactual scenarios, increasing their generalizability. Extensive experiments on various datasets show that our method effectively reduces hallucinations and improves the model's faithfulness to images, demonstrating high portability across both small-scale and large-scale image-to-text models. The code is available at https://github.com/Aman-4-Real/See-or-Guess.
Authors: Chao Wang, Neo Wu, Lin Ning, Luyang Liu, Jun Xie, Shawn O'Banion, Bradley Green
Abstract: Large language models (LLMs) have shown remarkable capabilities in generating user summaries from a long list of raw user activity data. These summaries capture essential user information such as preferences and interests, and therefore are invaluable for LLM-based personalization applications, such as explainable recommender systems. However, the development of new summarization techniques is hindered by the lack of ground-truth labels, the inherent subjectivity of user summaries, and human evaluation which is often costly and time-consuming. To address these challenges, we introduce \UserSumBench, a benchmark framework designed to facilitate iterative development of LLM-based summarization approaches. This framework offers two key components: (1) A reference-free summary quality metric. We show that this metric is effective and aligned with human preferences across three diverse datasets (MovieLens, Yelp and Amazon Review). (2) A novel robust summarization method that leverages time-hierarchical summarizer and self-critique verifier to produce high-quality summaries while eliminating hallucination. This method serves as a strong baseline for further innovation in summarization techniques.
Authors: Zhen Ye, Peiwen Sun, Jiahe Lei, Hongzhan Lin, Xu Tan, Zheqi Dai, Qiuqiang Kong, Jianyi Chen, Jiahao Pan, Qifeng Liu, Yike Guo, Wei Xue
Abstract: Recent advancements in audio generation have been significantly propelled by the capabilities of Large Language Models (LLMs). The existing research on audio LLM has primarily focused on enhancing the architecture and scale of audio language models, as well as leveraging larger datasets, and generally, acoustic codecs, such as EnCodec, are used for audio tokenization. However, these codecs were originally designed for audio compression, which may lead to suboptimal performance in the context of audio LLM. Our research aims to address the shortcomings of current audio LLM codecs, particularly their challenges in maintaining semantic integrity in generated audio. For instance, existing methods like VALL-E, which condition acoustic token generation on text transcriptions, often suffer from content inaccuracies and elevated word error rates (WER) due to semantic misinterpretations of acoustic tokens, resulting in word skipping and errors. To overcome these issues, we propose a straightforward yet effective approach called X-Codec. X-Codec incorporates semantic features from a pre-trained semantic encoder before the Residual Vector Quantization (RVQ) stage and introduces a semantic reconstruction loss after RVQ. By enhancing the semantic ability of the codec, X-Codec significantly reduces WER in speech synthesis tasks and extends these benefits to non-speech applications, including music and sound generation. Our experiments in text-to-speech, music continuation, and text-to-sound tasks demonstrate that integrating semantic information substantially improves the overall performance of language models in audio generation. Our code and demo are available (Demo: https://x-codec-audio.github.io Code: https://github.com/zhenye234/xcodec)
URLs: https://x-codec-audio.github.io, https://github.com/zhenye234/xcodec)
Authors: Rhui Dih Lee, Laura Wynter, Raghu Kiran Ganti
Abstract: We present a toolkit for creating low-cost Mixture-of-Domain-Experts (MOE) from trained models. The toolkit can be used for creating a mixture from models or from adapters. We perform extensive tests and offer guidance on defining the architecture of the resulting MOE using the toolkit. A public repository is available.
Authors: Ali Norouzifar, Humam Kourani, Marcus Dees, Wil van der Aalst
Abstract: Discovering good process models is essential for different process analysis tasks such as conformance checking and process improvements. Automated process discovery methods often overlook valuable domain knowledge. This knowledge, including insights from domain experts and detailed process documentation, remains largely untapped during process discovery. This paper leverages Large Language Models (LLMs) to integrate such knowledge directly into process discovery. We use rules derived from LLMs to guide model construction, ensuring alignment with both domain knowledge and actual process executions. By integrating LLMs, we create a bridge between process knowledge expressed in natural language and the discovery of robust process models, advancing process discovery methodologies significantly. To showcase the usability of our framework, we conducted a case study with the UWV employee insurance agency, demonstrating its practical benefits and effectiveness.
Authors: Nicholas Pochinkov, Ben Pasero, Skylar Shibayama
Abstract: The use of transformer-based models is growing rapidly throughout society. With this growth, it is important to understand how they work, and in particular, how the attention mechanisms represent concepts. Though there are many interpretability methods, many look at models through their neuronal activations, which are poorly understood. We describe different lenses through which to view neuron activations, and investigate the effectiveness in language models and vision transformers through various methods of neural ablation: zero ablation, mean ablation, activation resampling, and a novel approach we term 'peak ablation'. Through experimental analysis, we find that in different regimes and models, each method can offer the lowest degradation of model performance compared to other methods, with resampling usually causing the most significant performance deterioration. We make our code available at https://github.com/nickypro/investigating-ablation.
Authors: Nicholas Pochinkov, Thomas Jones, Mohammed Rashidur Rahman
Abstract: Transformer models are increasingly prevalent in various applications, yet our understanding of their internal workings remains limited. This paper investigates the modularity and task specialization of neurons within transformer architectures, focusing on both vision (ViT) and language (Mistral 7B) models. Using a combination of selective pruning and MoEfication clustering techniques, we analyze the overlap and specialization of neurons across different tasks and data subsets. Our findings reveal evidence of task-specific neuron clusters, with varying degrees of overlap between related tasks. We observe that neuron importance patterns persist to some extent even in randomly initialized models, suggesting an inherent structure that training refines. Additionally, we find that neuron clusters identified through MoEfication correspond more strongly to task-specific neurons in earlier and later layers of the models. This work contributes to a more nuanced understanding of transformer internals and offers insights into potential avenues for improving model interpretability and efficiency.
Authors: Gueter Josmy Faure, Jia-Fong Yeh, Min-Hung Chen, Hung-Ting Su, Winston H. Hsu, Shang-Hong Lai
Abstract: While existing research often treats long-form videos as extended short videos, we propose a novel approach that more accurately reflects human cognition. This paper introduces BREASE: BRidging Episodes And SEmantics for Long-Form Video Understanding, a model that simulates episodic memory accumulation to capture action sequences and reinforces them with semantic knowledge dispersed throughout the video. Our work makes two key contributions: First, we develop an Episodic COmpressor (ECO) that efficiently aggregates crucial representations from micro to semi-macro levels. Second, we propose a Semantics reTRiever (SeTR) that enhances these aggregated representations with semantic information by focusing on the broader context, dramatically reducing feature dimensionality while preserving relevant macro-level information. Extensive experiments demonstrate that BREASE achieves state-of-the-art performance across multiple long video understanding benchmarks in both zero-shot and fully-supervised settings. The project page and code are at: https://joslefaure.github.io/assets/html/hermes.html.
Authors: Tzu-Quan Lin, Hung-yi Lee, Hao Tang
Abstract: Self-supervised models have had great success in learning speech representations that can generalize to various downstream tasks. However, most self-supervised models require a large amount of compute and multiple GPUs to train, significantly hampering the development of self-supervised learning. In an attempt to reduce the computation of training, we revisit the training of HuBERT, a highly successful self-supervised model. We improve and simplify several key components, including the loss function, input representation, and training in multiple stages. Our model, MelHuBERT, is able to achieve favorable performance on phone recognition, speaker identification, and automatic speech recognition against HuBERT, while saving 31.2% of the pre-training time, or equivalently 33.5% MACs per one second speech. The code and pre-trained models are available in https://github.com/nervjack2/MelHuBERT.
Authors: Navid Madani, Rabiraj Bandyopadhyay, Briony Swire-Thompson, Michael Miller Yoder, Kenneth Joseph
Abstract: Social media platforms provide users with a profile description field, commonly known as a ``bio," where they can present themselves to the world. A growing literature shows that text in these bios can improve our understanding of online self-presentation and behavior, but existing work relies exclusively on keyword-based approaches to do so. We here propose and evaluate a suite of \hl{simple, effective, and theoretically motivated} approaches to embed bios in spaces that capture salient dimensions of social meaning, such as age and partisanship. We \hl{evaluate our methods on four tasks, showing that the strongest one out-performs several practical baselines.} We then show the utility of our method in helping understand associations between self-presentation and the sharing of URLs from low-quality news sites on Twitter\hl{, with a particular focus on explore the interactions between age and partisanship, and exploring the effects of self-presentations of religiosity}. Our work provides new tools to help computational social scientists make use of information in bios, and provides new insights into how misinformation sharing may be perceived on Twitter.
Authors: Boan Liu, Liang Ding, Li Shen, Keqin Peng, Yu Cao, Dazhao Cheng, Dacheng Tao
Abstract: The Mixture of Experts (MoE) has emerged as a highly successful technique in deep learning, based on the principle of divide-and-conquer to maximize model capacity without significant additional computational cost. Even in the era of large-scale language models (LLMs), MoE continues to play a crucial role, as some researchers have indicated that GPT-4 adopts the MoE structure to ensure diverse inference results. However, MoE is susceptible to performance degeneracy, particularly evident in the issues of imbalance and homogeneous representation among experts. While previous studies have extensively addressed the problem of imbalance, the challenge of homogeneous representation remains unresolved. In this study, we shed light on the homogeneous representation problem, wherein experts in the MoE fail to specialize and lack diversity, leading to frustratingly high similarities in their representations (up to 99\% in a well-performed MoE model). This problem restricts the expressive power of the MoE and, we argue, contradicts its original intention. To tackle this issue, we propose a straightforward yet highly effective solution: OMoE, an orthogonal expert optimizer. Additionally, we introduce an alternating training strategy that encourages each expert to update in a direction orthogonal to the subspace spanned by other experts. Our algorithm facilitates MoE training in two key ways: firstly, it explicitly enhances representation diversity, and secondly, it implicitly fosters interaction between experts during orthogonal weights computation. Through extensive experiments, we demonstrate that our proposed optimization algorithm significantly improves the performance of fine-tuning the MoE model on the GLUE benchmark, SuperGLUE benchmark, question-answering task, and name entity recognition tasks.
Authors: Jennifer Hu, Kyle Mahowald, Gary Lupyan, Anna Ivanova, Roger Levy
Abstract: Do large language models (LLMs) make human-like linguistic generalizations? Dentella et al. (2023) ("DGL") prompt several LLMs ("Is the following sentence grammatically correct in English?") to elicit grammaticality judgments of 80 English sentences, concluding that LLMs demonstrate a "yes-response bias" and a "failure to distinguish grammatical from ungrammatical sentences". We re-evaluate LLM performance using well-established practices and find that DGL's data in fact provide evidence for just how well LLMs capture human behaviors. Models not only achieve high accuracy overall, but also capture fine-grained variation in human linguistic judgments.
Authors: Shima Imani, Hamid Palangi
Abstract: Large Language Models (LLMs) have demonstrated impressive performance across a wide range of applications; however, assessing their reasoning capabilities remains a significant challenge. In this paper, we introduce a framework grounded in group and symmetry principles, which have played a crucial role in fields such as physics and mathematics, and offer another way to evaluate their capabilities. While the proposed framework is general, to showcase the benefits of employing these properties, we focus on arithmetic reasoning and investigate the performance of these models on four group properties: closure, identity, inverse, and associativity. Our findings reveal that LLMs studied in this work struggle to preserve group properties across different test regimes. In the closure test, we observe biases towards specific outputs and an abrupt degradation in their performance from 100% to 0% after a specific sequence length. They also perform poorly in the identity test, which represents adding irrelevant information in the context, and show sensitivity when subjected to inverse test, which examines the robustness of the model with respect to negation. In addition, we demonstrate that breaking down problems into smaller steps helps LLMs in the associativity test that we have conducted. To support these tests we have developed a synthetic dataset which will be released.
Authors: Zhitao He, Pengfei Cao, Chenhao Wang, Zhuoran Jin, Yubo Chen, Jiexin Xu, Huaijun Li, Xiaojian Jiang, Kang Liu, Jun Zhao
Abstract: With the development of deep learning, natural language processing technology has effectively improved the efficiency of various aspects of the traditional judicial industry. However, most current efforts focus on tasks within individual judicial stages, making it difficult to handle complex tasks that span multiple stages. As the autonomous agents powered by large language models are becoming increasingly smart and able to make complex decisions in real-world settings, offering new insights for judicial intelligence. In this paper, (1) we propose a novel multi-agent framework, AgentsCourt, for judicial decision-making. Our framework follows the classic court trial process, consisting of court debate simulation, legal resources retrieval and decision-making refinement to simulate the decision-making of judge. (2) we introduce SimuCourt, a judicial benchmark that encompasses 420 Chinese judgment documents, spanning the three most common types of judicial cases. Furthermore, to support this task, we construct a large-scale legal knowledge base, Legal-KB, with multi-resource legal knowledge. (3) Extensive experiments show that our framework outperforms the existing advanced methods in various aspects, especially in generating legal articles, where our model achieves significant improvements of 8.6% and 9.1% F1 score in the first and second instance settings, respectively.
Authors: Jun-En Ding, Phan Nguyen Minh Thao, Wen-Chih Peng, Jian-Zhe Wang, Chun-Cheng Chug, Min-Chen Hsieh, Yun-Chien Tseng, Ling Chen, Dongsheng Luo, Chi-Te Wang, Pei-fu Chen, Feng Liu, Fang-Ming Hung
Abstract: Chronic diseases such as diabetes are the leading causes of morbidity and mortality worldwide. Numerous research studies have been attempted with various deep learning models in diagnosis. However, most previous studies had certain limitations, including using publicly available datasets (e.g. MIMIC), and imbalanced data. In this study, we collected five-year electronic health records (EHRs) from the Taiwan hospital database, including 1,420,596 clinical notes, 387,392 laboratory test results, and more than 1,505 laboratory test items, focusing on research pre-training large language models. We proposed a novel Large Language Multimodal Models (LLMMs) framework incorporating multimodal data from clinical notes and laboratory test results for the prediction of chronic disease risk. Our method combined a text embedding encoder and multi-head attention layer to learn laboratory test values, utilizing a deep neural network (DNN) module to merge blood features with chronic disease semantics into a latent space. In our experiments, we observe that clinicalBERT and PubMed-BERT, when combined with attention fusion, can achieve an accuracy of 73% in multiclass chronic diseases and diabetes prediction. By transforming laboratory test values into textual descriptions and employing the Flan T-5 model, we achieved a 76% Area Under the ROC Curve (AUROC), demonstrating the effectiveness of leveraging numerical text data for training and inference in language models. This approach significantly improves the accuracy of early-stage diabetes prediction.
Authors: Ramon Abilio, Guilherme Palermo Coelho, Ana Estela Antunes da Silva
Abstract: Since 2018, when the Transformer architecture was introduced, Natural Language Processing has gained significant momentum with pre-trained Transformer-based models that can be fine-tuned for various tasks. Most models are pre-trained on large English corpora, making them less applicable to other languages, such as Brazilian Portuguese. In our research, we identified two models pre-trained in Brazilian Portuguese (BERTimbau and PTT5) and two multilingual models (mBERT and mT5). BERTimbau and mBERT use only the Encoder module, while PTT5 and mT5 use both the Encoder and Decoder. Our study aimed to evaluate their performance on a financial Named Entity Recognition (NER) task and determine the computational requirements for fine-tuning and inference. To this end, we developed the Brazilian Financial NER (BraFiNER) dataset, comprising sentences from Brazilian banks' earnings calls transcripts annotated using a weakly supervised approach. Additionally, we introduced a novel approach that reframes the token classification task as a text generation problem. After fine-tuning the models, we evaluated them using performance and error metrics. Our findings reveal that BERT-based models consistently outperform T5-based models. While the multilingual models exhibit comparable macro F1-scores, BERTimbau demonstrates superior performance over PTT5. In terms of error metrics, BERTimbau outperforms the other models. We also observed that PTT5 and mT5 generated sentences with changes in monetary and percentage values, highlighting the importance of accuracy and consistency in the financial domain. Our findings provide insights into the differing performance of BERT- and T5-based models for the NER task.
Authors: Vivek Khetan
Abstract: This position paper proposes a systematic approach towards developing a framework to help select the most effective embedding models for natural language processing (NLP) tasks, addressing the challenge posed by the proliferation of both proprietary and open-source encoder models.
Authors: Yongcheng Zeng, Guoqing Liu, Weiyu Ma, Ning Yang, Haifeng Zhang, Jun Wang
Abstract: Fine-tuning pre-trained Large Language Models (LLMs) is essential to align them with human values and intentions. This process often utilizes methods like pairwise comparisons and KL divergence against a reference LLM, focusing on the evaluation of full answers generated by the models. However, the generation of these responses occurs in a token level, following a sequential, auto-regressive fashion. In this paper, we introduce Token-level Direct Preference Optimization (TDPO), a novel approach to align LLMs with human preferences by optimizing policy at the token level. Unlike previous methods, which face challenges in divergence efficiency, TDPO incorporates forward KL divergence constraints for each token, improving alignment and diversity. Utilizing the Bradley-Terry model for a token-based reward system, TDPO enhances the regulation of KL divergence, while preserving simplicity without the need for explicit reward modeling. Experimental results across various text tasks demonstrate TDPO's superior performance in balancing alignment with generation diversity. Notably, fine-tuning with TDPO strikes a better balance than DPO in the controlled sentiment generation and single-turn dialogue datasets, and significantly improves the quality of generated responses compared to both DPO and PPO-based RLHF methods. Our code is open-sourced at https://github.com/Vance0124/Token-level-Direct-Preference-Optimization.
URLs: https://github.com/Vance0124/Token-level-Direct-Preference-Optimization.
Authors: Hsuvas Borkakoty, Luis Espinosa-Anke
Abstract: Hoaxes are a recognised form of disinformation created deliberately, with potential serious implications in the credibility of reference knowledge resources such as Wikipedia. What makes detecting Wikipedia hoaxes hard is that they often are written according to the official style guidelines. In this work, we first provide a systematic analysis of similarities and discrepancies between legitimate and hoax Wikipedia articles, and introduce Hoaxpedia, a collection of 311 hoax articles (from existing literature and official Wikipedia lists), together with semantically similar legitimate articles, which together form a binary text classification dataset aimed at fostering research in automated hoax detection. In this paper, We report results after analyzing several language models, hoax-to-legit ratios, and the amount of text classifiers are exposed to (full article vs the article's definition alone). Our results suggest that detecting deceitful content in Wikipedia based on content alone is hard but feasible, and complement our analysis with a study on the differences in distributions in edit histories, and find that looking at this feature yields better classification results than context.
Authors: Rabindra Lamsal, Maria Rodriguez Read, Shanika Karunasekera, Muhammad Imran
Abstract: During times of crisis, social media platforms play a crucial role in facilitating communication and coordinating resources. In the midst of chaos and uncertainty, communities often rely on these platforms to share urgent pleas for help, extend support, and organize relief efforts. However, the overwhelming volume of conversations during such periods can escalate to unprecedented levels, necessitating the automated identification and matching of requests and offers to streamline relief operations. Additionally, there is a notable absence of studies conducted in multi-lingual settings, despite the fact that any geographical area can have a diverse linguistic population. Therefore, we propose CReMa (Crisis Response Matcher), a systematic approach that integrates textual, temporal, and spatial features to address the challenges of effectively identifying and matching requests and offers on social media platforms during emergencies. Our approach utilizes a crisis-specific pre-trained model and a multi-lingual embedding space. We emulate human decision-making to compute temporal and spatial features and non-linearly weigh the textual features. The results from our experiments are promising, outperforming strong baselines. Additionally, we introduce a novel multi-lingual dataset simulating help-seeking and offering assistance on social media in 16 languages and conduct comprehensive cross-lingual experiments. Furthermore, we analyze a million-scale geotagged global dataset to understand patterns in seeking help and offering assistance on social media. Overall, these contributions advance the field of crisis informatics and provide benchmarks for future research in the area.
Authors: Vatsal Raina, Mark Gales
Abstract: Enterprise retrieval augmented generation (RAG) offers a highly flexible framework for combining powerful large language models (LLMs) with internal, possibly temporally changing, documents. In RAG, documents are first chunked. Relevant chunks are then retrieved for a user query, which are passed as context to a synthesizer LLM to generate the query response. However, the retrieval step can limit performance, as incorrect chunks can lead the synthesizer LLM to generate a false response. This work applies a zero-shot adaptation of standard dense retrieval steps for more accurate chunk recall. Specifically, a chunk is first decomposed into atomic statements. A set of synthetic questions are then generated on these atoms (with the chunk as the context). Dense retrieval involves finding the closest set of synthetic questions, and associated chunks, to the user query. It is found that retrieval with the atoms leads to higher recall than retrieval with chunks. Further performance gain is observed with retrieval using the synthetic questions generated over the atoms. Higher recall at the retrieval step enables higher performance of the enterprise LLM using the RAG pipeline.
Authors: Jonas Becker, Jan Philip Wahle, Bela Gipp, Terry Ruas
Abstract: Text generation has become more accessible than ever, and the increasing interest in these systems, especially those using large language models, has spurred an increasing number of related publications. We provide a systematic literature review comprising 244 selected papers between 2017 and 2024. This review categorizes works in text generation into five main tasks: open-ended text generation, summarization, translation, paraphrasing, and question answering. For each task, we review their relevant characteristics, sub-tasks, and specific challenges (e.g., missing datasets for multi-document summarization, coherence in story generation, and complex reasoning for question answering). Additionally, we assess current approaches for evaluating text generation systems and ascertain problems with current metrics. Our investigation shows nine prominent challenges common to all tasks and sub-tasks in recent text generation publications: bias, reasoning, hallucinations, misuse, privacy, interpretability, transparency, datasets, and computing. We provide a detailed analysis of these challenges, their potential solutions, and which gaps still require further engagement from the community. This systematic literature review targets two main audiences: early career researchers in natural language processing looking for an overview of the field and promising research directions, as well as experienced researchers seeking a detailed view of tasks, evaluation methodologies, open challenges, and recent mitigation strategies.
Authors: Jing Li, Zhijie Sun, Dachao Lin, Xuan He, Yi Lin, Binfan Zheng, Li Zeng, Rongqian Zhao, Xin Chen
Abstract: Mixture-of-Experts (MoE) architectures have emerged as a paradigm-shifting approach for large language models (LLMs), offering unprecedented computational efficiency. However, these architectures grapple with challenges of token distribution imbalance and expert homogenization, impeding optimal semantic generalization. We introduce a novel framework that redefines MoE routing through affinity-driven active selection. The innovations for the framework encompass: (1) A rigorous formulation of expert-token affinity metrics. (2) An adaptive bidirectional selection mechanism leveraging resonance between experts and tokens. (3) Theoretical derivation and experimental evidence of reduced expert capacity bounds under dynamic token distribution evolution. It is also integrated with orthogonal feature extraction module and an optimized loss function for expert localization. Our theoretical analysis demonstrates that this approach mitigates expert homogenization while enabling substantial capacity boundary reduction. Experimental validation corroborates these findings: it achieves a 40% reduction in token processed by each expert without compromising model convergence or efficacy. When coupled with communication optimizations, the training efficiency improvements of 5.4% to 46.6% can be observed. After supervised fine-tuning, it exhibits performance gains of 9.7% to 14.1% across GDAD, C-Eval, and TeleQnA benchmarks.
Authors: Neelabh Sinha, Vinija Jain, Aman Chadha
Abstract: The rapid rise of Language Models (LMs) has expanded their use in several applications. Yet, due to constraints of model size, associated cost, or proprietary restrictions, utilizing state-of-the-art (SOTA) LLMs is not always feasible. With open, smaller LMs emerging, more applications can leverage their capabilities, but selecting the right LM can be challenging as smaller LMs don't perform well universally. This work tries to bridge this gap by proposing a framework to experimentally evaluate small, open LMs in practical settings through measuring semantic correctness of outputs across three practical aspects: task types, application domains and reasoning types, using diverse prompt styles. It also conducts an in-depth comparison of 10 small, open LMs to identify best LM and prompt style depending on specific application requirement using the proposed framework. We also show that if selected appropriately, they can outperform SOTA LLMs like DeepSeek-v2, GPT-4o-mini, Gemini-1.5-Pro, and even compete with GPT-4o.
Authors: Jo\~ao A. Leite, Olesya Razuvayevskaya, Kalina Bontcheva, Carolina Scarton
Abstract: This work introduces EUvsDisinfo, a multilingual dataset of disinformation articles originating from pro-Kremlin outlets, along with trustworthy articles from credible / less biased sources. It is sourced directly from the debunk articles written by experts leading the EUvsDisinfo project. Our dataset is the largest to-date resource in terms of the overall number of articles and distinct languages. It also provides the largest topical and temporal coverage. Using this dataset, we investigate the dissemination of pro-Kremlin disinformation across different languages, uncovering language-specific patterns targeting certain disinformation topics. We further analyse the evolution of topic distribution over an eight-year period, noting a significant surge in disinformation content before the full-scale invasion of Ukraine in 2022. Lastly, we demonstrate the dataset's applicability in training models to effectively distinguish between disinformation and trustworthy content in multilingual settings.
Authors: Yuetong Zhao, Hongyu Cao, Xianyu Zhao, Zhijian Ou
Abstract: Since the launch of ChatGPT at the end of 2022, generative dialogue models represented by ChatGPT have quickly become essential tools in daily life. As user expectations increase, enhancing the capability of generative dialogue models to solve complex problems has become a focal point of current research. This paper delves into the effectiveness of the RAFT (Retrieval Augmented Fine-Tuning) method in improving the performance of Generative dialogue models. RAFT combines chain-of-thought with model supervised fine-tuning (SFT) and retrieval augmented generation (RAG), which significantly enhanced the model's information extraction and logical reasoning abilities. We evaluated the RAFT method across multiple datasets and analysed its performance in various reasoning tasks, including long-form QA and short-form QA tasks, tasks in both Chinese and English, and supportive and comparison reasoning tasks. Notably, it addresses the gaps in previous research regarding long-form QA tasks and Chinese datasets. Moreover, we also evaluate the benefit of the chain-of-thought (CoT) in the RAFT method. This work offers valuable insights for studies focused on enhancing the performance of generative dialogue models.
Authors: Shanbo Cheng, Zhichao Huang, Tom Ko, Hang Li, Ningxin Peng, Lu Xu, Qini Zhang
Abstract: In this paper, we present Cross Language Agent -- Simultaneous Interpretation, CLASI, a high-quality and human-like Simultaneous Speech Translation (SiST) System. Inspired by professional human interpreters, we utilize a novel data-driven read-write strategy to balance the translation quality and latency. To address the challenge of translating in-domain terminologies, CLASI employs a multi-modal retrieving module to obtain relevant information to augment the translation. Supported by LLMs, our approach can generate error-tolerated translation by considering the input audio, historical context, and retrieved information. Experimental results show that our system outperforms other systems by significant margins. Aligned with professional human interpreters, we evaluate CLASI with a better human evaluation metric, valid information proportion (VIP), which measures the amount of information that can be successfully conveyed to the listeners. In the real-world scenarios, where the speeches are often disfluent, informal, and unclear, CLASI achieves VIP of 81.3% and 78.0% for Chinese-to-English and English-to-Chinese translation directions, respectively. In contrast, state-of-the-art commercial or open-source systems only achieve 35.4% and 41.6%. On the extremely hard dataset, where other systems achieve under 13% VIP, CLASI can still achieve 70% VIP.
Authors: Yujie Feng, Xu Chu, Yongxin Xu, Zexin Lu, Bo Liu, Philip S. Yu, Xiao-Ming Wu
Abstract: Language model continual learning (CL) has recently attracted significant interest for its ability to adapt large language models (LLMs) to dynamic real-world scenarios without retraining. A major challenge in this domain is catastrophic forgetting, where models lose previously acquired knowledge upon learning new tasks. Existing approaches commonly utilize multiple parameter-efficient fine-tuning (PEFT) blocks to acquire task-specific knowledge, yet these methods are inefficient and fail to leverage potential knowledge transfer across tasks. In this paper, we introduce a novel CL framework for language models, named Task Skill Localization and Consolidation (TaSL), which boosts knowledge transfer without depending on memory replay. TaSL initially segregates the model into 'skill units' based on parameter dependencies, allowing for more precise control. Subsequently, it employs a novel group-wise skill localization technique to ascertain the importance distribution of skill units for a new task. By comparing this importance distribution with those from previous tasks, we implement a fine-grained skill consolidation strategy that retains task-specific knowledge, thereby preventing forgetting, and updates task-shared knowledge, which facilitates bi-directional knowledge transfer. As a result, TaSL achieves an optimal balance between retaining prior knowledge and excelling in new tasks. TaSL also demonstrates strong generalizability, making it suitable for various base models and adaptable to PEFT methods like LoRA. Furthermore, it offers notable extensibility, supporting enhancements through integration with memory replay techniques. Comprehensive experiments conducted on two CL benchmarks, involving models ranging from 220M to 7B parameters, affirm the effectiveness of TaSL and its variants across different settings.
Authors: Christoph Auer, Maksym Lysak, Ahmed Nassar, Michele Dolfi, Nikolaos Livathinos, Panos Vagenas, Cesar Berrospi Ramis, Matteo Omenetti, Fabian Lindlbauer, Kasper Dinkla, Lokesh Mishra, Yusik Kim, Shubham Gupta, Rafael Teixeira de Lima, Valery Weber, Lucas Morin, Ingmar Meijer, Viktor Kuropiatnyk, Peter W. J. Staar
Abstract: This technical report introduces Docling, an easy to use, self-contained, MIT-licensed open-source package for PDF document conversion. It is powered by state-of-the-art specialized AI models for layout analysis (DocLayNet) and table structure recognition (TableFormer), and runs efficiently on commodity hardware in a small resource budget. The code interface allows for easy extensibility and addition of new features and models.
Authors: Li Du, Zhouhao Sun, Xiao Ding, Yixuan Ma, Yang Zhao, Kaitao Qiu, Ting Liu, Bing Qin
Abstract: Although achieving promising performance, recent analyses show that current generative large language models (LLMs) may still capture dataset biases and utilize them for generation, leading to poor generalizability and harmfulness of LLMs. However, due to the diversity of dataset biases and the over-optimization problem, previous prior-knowledge-based debiasing methods and fine-tuning-based debiasing methods may not be suitable for current LLMs. To address this issue, we explore combining active learning with the causal mechanisms and propose a casual-guided active learning (CAL) framework, which utilizes LLMs itself to automatically and autonomously identify informative biased samples and induce the bias patterns. Then a cost-effective and efficient in-context learning based method is employed to prevent LLMs from utilizing dataset biases during generation. Experimental results show that CAL can effectively recognize typical biased instances and induce various bias patterns for debiasing LLMs.
Authors: Haley Lepp, Parth Sarin
Abstract: In this provocation, we discuss the English dominance of the AI research community, arguing that the requirement for English language publishing upholds and reinforces broader regimes of extraction in AI. While large language models and machine translation have been celebrated as a way to break down barriers, we regard their use as a symptom of linguistic exclusion of scientists and potential readers. We propose alternative futures for a healthier publishing culture, organized around three themes: administering conferences in the languages of the country in which they are held, instructing peer reviewers not to adjudicate the language appropriateness of papers, and offering opportunities to publish and present in multiple languages. We welcome new translations of this piece. Please contact the authors if you would like to contribute one.
Authors: Adamu Lawan, Juhua Pu, Haruna Yunusa, Muhammad Lawan, Aliyu Umar, Adamu Sani Yahya
Abstract: Multimodal aspect-based sentiment analysis (MABSA) enhances sentiment detection by combining text with other data types like images. However, despite setting significant benchmarks, attention mechanisms exhibit limitations in efficiently modelling long-range dependencies between aspect and opinion targets within the text. They also face challenges in capturing global-context dependencies for visual representations. To this end, we propose Kolmogorov-Arnold Networks (KANs) and Selective State Space model (Mamba) transformer (DualKanbaFormer), a novel architecture to address the above issues. We leverage the power of Mamba to capture global context dependencies, Multi-head Attention (MHA) to capture local context dependencies, and KANs to capture non-linear modelling patterns for both textual representations (textual KanbaFormer) and visual representations (visual KanbaFormer). Furthermore, we fuse the textual KanbaFormer and visual KanbaFomer with a gated fusion layer to capture the inter-modality dynamics. According to extensive experimental results, our model outperforms some state-of-the-art (SOTA) studies on two public datasets.
Authors: Martha Lewis, Nihal V. Nayak, Peilin Yu, Qinan Yu, Jack Merullo, Stephen H. Bach, Ellie Pavlick
Abstract: Large-scale neural network models combining text and images have made incredible progress in recent years. However, it remains an open question to what extent such models encode compositional representations of the concepts over which they operate, such as correctly identifying "red cube" by reasoning over the constituents "red" and "cube". In this work, we focus on the ability of a large pretrained vision and language model (CLIP) to encode compositional concepts and to bind variables in a structure-sensitive way (e.g., differentiating "cube behind sphere" from "sphere behind cube"). To inspect the performance of CLIP, we compare several architectures from research on compositional distributional semantics models (CDSMs), a line of research that attempts to implement traditional compositional linguistic structures within embedding spaces. We benchmark them on three synthetic datasets - single-object, two-object, and relational - designed to test concept binding. We find that CLIP can compose concepts in a single-object setting, but in situations where concept binding is needed, performance drops dramatically. At the same time, CDSMs also perform poorly, with best performance at chance level.
Authors: Yixiao Zhang, Akira Maezawa, Gus Xia, Kazuhiko Yamamoto, Simon Dixon
Abstract: Creating music is iterative, requiring varied methods at each stage. However, existing AI music systems fall short in orchestrating multiple subsystems for diverse needs. To address this gap, we introduce Loop Copilot, a novel system that enables users to generate and iteratively refine music through an interactive, multi-round dialogue interface. The system uses a large language model to interpret user intentions and select appropriate AI models for task execution. Each backend model is specialized for a specific task, and their outputs are aggregated to meet the user's requirements. To ensure musical coherence, essential attributes are maintained in a centralized table. We evaluate the effectiveness of the proposed system through semi-structured interviews and questionnaires, highlighting its utility not only in facilitating music creation but also its potential for broader applications.
Authors: Xiaoxu Xu, Yitian Yuan, Qiudan Zhang, Wenhui Wu, Zequn Jie, Lin Ma, Xu Wang
Abstract: Learning to ground natural language queries to target objects or regions in 3D point clouds is quite essential for 3D scene understanding. Nevertheless, existing 3D visual grounding approaches require a substantial number of bounding box annotations for text queries, which is time-consuming and labor-intensive to obtain. In this paper, we propose 3D-VLA, a weakly supervised approach for 3D visual grounding based on Visual Linguistic Alignment. Our 3D-VLA exploits the superior ability of current large-scale vision-language models (VLMs) on aligning the semantics between texts and 2D images, as well as the naturally existing correspondences between 2D images and 3D point clouds, and thus implicitly constructs correspondences between texts and 3D point clouds with no need for fine-grained box annotations in the training procedure. During the inference stage, the learned text-3D correspondence will help us ground the text queries to the 3D target objects even without 2D images. To the best of our knowledge, this is the first work to investigate 3D visual grounding in a weakly supervised manner by involving large scale vision-language models, and extensive experiments on ReferIt3D and ScanRefer datasets demonstrate that our 3D-VLA achieves comparable and even superior results over the fully supervised methods.
Authors: Hui Zong, Rongrong Wu, Jiaxue Cha, Weizhe Feng, Erman Wu, Jiakun Li, Aibin Shao, Liang Tao, Zuofeng Li, Buzhou Tang, Bairong Shen
Abstract: Objective: This study aims to review the recent advances in community challenges for biomedical text mining in China. Methods: We collected information of evaluation tasks released in community challenges of biomedical text mining, including task description, dataset description, data source, task type and related links. A systematic summary and comparative analysis were conducted on various biomedical natural language processing tasks, such as named entity recognition, entity normalization, attribute extraction, relation extraction, event extraction, text classification, text similarity, knowledge graph construction, question answering, text generation, and large language model evaluation. Results: We identified 39 evaluation tasks from 6 community challenges that spanned from 2017 to 2023. Our analysis revealed the diverse range of evaluation task types and data sources in biomedical text mining. We explored the potential clinical applications of these community challenge tasks from a translational biomedical informatics perspective. We compared with their English counterparts, and discussed the contributions, limitations, lessons and guidelines of these community challenges, while highlighting future directions in the era of large language models. Conclusion: Community challenge evaluation competitions have played a crucial role in promoting technology innovation and fostering interdisciplinary collaboration in the field of biomedical text mining. These challenges provide valuable platforms for researchers to develop state-of-the-art solutions.
Authors: Sowmya S. Sundaram, Benjamin Solomon, Avani Khatri, Anisha Laumas, Purvesh Khatri, Mark A. Musen
Abstract: Metadata play a crucial role in ensuring the findability, accessibility, interoperability, and reusability of datasets. This paper investigates the potential of large language models (LLMs), specifically GPT-4, to improve adherence to metadata standards. We conducted experiments on 200 random data records describing human samples relating to lung cancer from the NCBI BioSample repository, evaluating GPT-4's ability to suggest edits for adherence to metadata standards. We computed the adherence accuracy of field name-field value pairs through a peer review process, and we observed a marginal average improvement in adherence to the standard data dictionary from 79% to 80% (p<0.5). We then prompted GPT-4 with domain information in the form of the textual descriptions of CEDAR templates and recorded a significant improvement to 97% from 79% (p<0.01). These results indicate that, while LLMs may not be able to correct legacy metadata to ensure satisfactory adherence to standards when unaided, they do show promise for use in automated metadata curation when integrated with a structured knowledge base
Authors: Yupeng Cao, Zhi Chen, Qingyun Pei, Nathan Jinseok Lee, K. P. Subbalakshmi, Papa Momar Ndiaye
Abstract: In the realm of financial analytics, leveraging unstructured data, such as earnings conference calls (ECCs), to forecast stock volatility is a critical challenge that has attracted both academics and investors. While previous studies have used multimodal deep learning-based models to obtain a general view of ECCs for volatility predicting, they often fail to capture detailed, complex information. Our research introduces a novel framework: \textbf{ECC Analyzer}, which utilizes large language models (LLMs) to extract richer, more predictive content from ECCs to aid the model's prediction performance. We use the pre-trained large models to extract textual and audio features from ECCs and implement a hierarchical information extraction strategy to extract more fine-grained information. This strategy first extracts paragraph-level general information by summarizing the text and then extracts fine-grained focus sentences using Retrieval-Augmented Generation (RAG). These features are then fused through multimodal feature fusion to perform volatility prediction. Experimental results demonstrate that our model outperforms traditional analytical benchmarks, confirming the effectiveness of advanced LLM techniques in financial analysis.
Authors: Yui Sudo, Yosuke Fukumoto, Muhammad Shakeel, Yifan Peng, Shinji Watanabe
Abstract: Deep biasing (DB) enhances the performance of end-to-end automatic speech recognition (E2E-ASR) models for rare words or contextual phrases using a bias list. However, most existing methods treat bias phrases as sequences of subwords in a predefined static vocabulary. This naive sequence decomposition produces unnatural token patterns, significantly lowering their occurrence probability. More advanced techniques address this problem by expanding the vocabulary with additional modules, including the external language model shallow fusion or rescoring. However, they result in increasing the workload due to the additional modules. This paper proposes a dynamic vocabulary where bias tokens can be added during inference. Each entry in a bias list is represented as a single token, unlike a sequence of existing subword tokens. This approach eliminates the need to learn subword dependencies within the bias phrases. This method is easily applied to various architectures because it only expands the embedding and output layers in common E2E-ASR architectures. Experimental results demonstrate that the proposed method improves the bias phrase WER on English and Japanese datasets by 3.1 -- 4.9 points compared with the conventional DB method.
Authors: Mehant Kammakomati, Sameer Pimparkhede, Srikanth Tamilselvam, Prince Kumar, Pushpak Bhattacharyya
Abstract: Recent work shows Large Language Models (LLMs) struggle to understand natural language constraints for various text generation tasks in zero- and few-shot settings. While, in the code domain, there is wide usage of constraints in code format to maintain the integrity of code written in Domain-Specific Languages (DSLs) like JSON and YAML which are widely used for system-level programming tasks in enterprises. Given that LLMs are increasingly used for system-level code tasks, evaluating if they can comprehend these code constraints is crucial. However, no work has been done to evaluate their controllability over code constraints. Hence, we introduce ConCodeEval, a first-of-its-kind benchmark having two novel tasks for code constraints across five representations. Our findings suggest that language models struggle with code constraints. Code languages that perform excellently for normal code tasks do not perform well when the same languages represent fine-grained constraints.
Authors: Sibo Yi, Yule Liu, Zhen Sun, Tianshuo Cong, Xinlei He, Jiaxing Song, Ke Xu, Qi Li
Abstract: Large Language Models (LLMs) have performed exceptionally in various text-generative tasks, including question answering, translation, code completion, etc. However, the over-assistance of LLMs has raised the challenge of "jailbreaking", which induces the model to generate malicious responses against the usage policy and society by designing adversarial prompts. With the emergence of jailbreak attack methods exploiting different vulnerabilities in LLMs, the corresponding safety alignment measures are also evolving. In this paper, we propose a comprehensive and detailed taxonomy of jailbreak attack and defense methods. For instance, the attack methods are divided into black-box and white-box attacks based on the transparency of the target model. Meanwhile, we classify defense methods into prompt-level and model-level defenses. Additionally, we further subdivide these attack and defense methods into distinct sub-classes and present a coherent diagram illustrating their relationships. We also conduct an investigation into the current evaluation methods and compare them from different perspectives. Our findings aim to inspire future research and practical implementations in safeguarding LLMs against adversarial attacks. Above all, although jailbreak remains a significant concern within the community, we believe that our work enhances the understanding of this domain and provides a foundation for developing more secure LLMs.
Authors: Amey Agrawal, Anmol Agarwal, Nitin Kedia, Jayashree Mohan, Souvik Kundu, Nipun Kwatra, Ramachandran Ramjee, Alexey Tumanov
Abstract: Serving large language models (LLMs) in production can incur substantial costs, which has prompted recent advances in inference system optimizations. Today, these systems are evaluated against conventional latency and throughput metrics (eg. TTFT, TBT, Normalised Latency and TPOT). However, these metrics fail to fully capture the nuances of LLM inference, leading to an incomplete assessment of user-facing performance crucial for real-time applications such as chat and translation. In this paper, we first identify the pitfalls of current performance metrics in evaluating LLM inference systems. We then propose Etalon, a comprehensive performance evaluation framework that includes fluidity-index -- a novel metric designed to reflect the intricacies of the LLM inference process and its impact on real-time user experience. Finally, we evaluate various existing open-source platforms and model-as-a-service offerings using Etalon, discussing their strengths and weaknesses. Etalon is available at https://github.com/project-etalon/etalon.
Authors: Karel D'Oosterlinck, Winnie Xu, Chris Develder, Thomas Demeester, Amanpreet Singh, Christopher Potts, Douwe Kiela, Shikib Mehri
Abstract: Large Language Models (LLMs) are often aligned using contrastive alignment objectives and preference pair datasets. The interaction between model, paired data, and objective makes alignment a complicated procedure, sometimes producing subpar results. We study this and find that (i) preference data gives a better learning signal when the underlying responses are contrastive, and (ii) alignment objectives lead to better performance when they specify more control over the model during training. Based on these insights, we introduce Contrastive Learning from AI Revisions (CLAIR), a data-creation method which leads to more contrastive preference pairs, and Anchored Preference Optimization (APO), a controllable and more stable alignment objective. We align Llama-3-8B-Instruct using various comparable datasets and alignment objectives and measure MixEval-Hard scores, which correlate highly with human judgments. The CLAIR preferences lead to the strongest performance out of all datasets, and APO consistently outperforms less controllable objectives. Our best model, trained on 32K CLAIR preferences with APO, improves Llama-3-8B-Instruct by 7.65%, closing the gap with GPT4-turbo by 45%. Our code is available at https://github.com/ContextualAI/CLAIR_and_APO.
Authors: Sihang Li, Jin Huang, Jiaxi Zhuang, Yaorui Shi, Xiaochen Cai, Mingjun Xu, Xiang Wang, Linfeng Zhang, Guolin Ke, Hengxing Cai
Abstract: Scientific literature understanding is crucial for extracting targeted information and garnering insights, thereby significantly advancing scientific discovery. Despite the remarkable success of Large Language Models (LLMs), they face challenges in scientific literature understanding, primarily due to (1) a lack of scientific knowledge and (2) unfamiliarity with specialized scientific tasks. To develop an LLM specialized in scientific literature understanding, we propose a hybrid strategy that integrates continual pre-training (CPT) and supervised fine-tuning (SFT), to simultaneously infuse scientific domain knowledge and enhance instruction-following capabilities for domain-specific tasks.cIn this process, we identify two key challenges: (1) constructing high-quality CPT corpora, and (2) generating diverse SFT instructions. We address these challenges through a meticulous pipeline, including PDF text extraction, parsing content error correction, quality filtering, and synthetic instruction creation. Applying this strategy, we present a suite of LLMs: SciLitLLM, specialized in scientific literature understanding. These models demonstrate promising performance on scientific literature understanding benchmarks. Our contributions are threefold: (1) We present an effective framework that integrates CPT and SFT to adapt LLMs to scientific literature understanding, which can also be easily adapted to other domains. (2) We propose an LLM-based synthesis method to generate diverse and high-quality scientific instructions, resulting in a new instruction set -- SciLitIns -- for supervised fine-tuning in less-represented scientific domains. (3) SciLitLLM achieves promising performance improvements on scientific literature understanding benchmarks.
Authors: Zhifei Xie, Changqiao Wu
Abstract: Recent advances in language models have achieved significant progress. GPT-4o, as a new milestone, has enabled real-time conversations with humans, demonstrating near-human natural fluency. Such human-computer interaction necessitates models with the capability to perform reasoning directly with the audio modality and generate output in streaming. However, this remains beyond the reach of current academic models, as they typically depend on extra TTS systems for speech synthesis, resulting in undesirable latency. This paper introduces the Mini-Omni, an audio-based end-to-end conversational model, capable of real-time speech interaction. To achieve this capability, we propose a text-instructed speech generation method, along with batch-parallel strategies during inference to further boost the performance. Our method also helps to retain the original model's language capabilities with minimal degradation, enabling other works to establish real-time interaction capabilities. We call this training method "Any Model Can Talk". We also introduce the VoiceAssistant-400K dataset to fine-tune models optimized for speech output. To our best knowledge, Mini-Omni is the first fully end-to-end, open-source model for real-time speech interaction, offering valuable potential for future research.