EMNLP 2024 论文列表

Large language models (LLMs) can perform complex reasoning by generating intermediate reasoning steps using chain-of-thought prompting under zero-shot or few-shot settings. However, zero-shot prompting always encounters low performance, and the superior performance of few-shot prompting hinges on the manual-crafting of task-specific demonstrations one by one. In this paper, we present **RoSE** (**R**easoning with **O**rchestrated **S**treaming **E**xperiences), a general framework for solving reasoning tasks that can self-improve as it answers various reasoning questions. To enable RoSE, we describe an architecture that extends an LLM to store all answered reasoning questions and their reasoning steps in a streaming experience pool and orchestrate helpful questions from the pool to assist itself in answering new questions. To set up a question-aware orchestration mechanism, RoSE first calculates the similarity of each question in the pool with the question to be answered. Since the solution to each question in the experience pool is not always correct, RoSE will sort the questions according to their similarity with the question to be answered, and then uniformly divide them into multiple buckets. It finally extracts one question from each bucket to make the extracted questions more diverse. To make the extracted questions help RoSE answer new questions as much as possible, we introduce two other attributes of uncertainty and complexity for each question. RoSE will preferentially select the questions with low uncertainty and high complexity from each bucket. We evaluate the versatility of RoSE in various complex reasoning tasks and LLMs, such as arithmetic and commonsense reasoning, and find that it can achieve excellent performance without any labeled data and pre-set unlabeled data.

Existing speculative decoding methods typically require additional model structure and training processes to assist the model for draft token generation. This makes the migration of acceleration methods to the new model more costly and more demanding on device memory. To address this problem, we propose the Make Some Noise (MSN) training framework as a replacement for the supervised fine-tuning stage of the large language model. The training method simply introduces some noise at the input for the model to learn the denoising task. It significantly enhances the parallel decoding capability of the model without affecting the original task capability. In addition, we propose a tree-based retrieval-augmented Jacobi (TR-Jacobi) decoding strategy to further improve the inference speed of MSN models. Experiments in both the general and code domains have shown that MSN can improve inference speed by 2.3-2.7x times without compromising model performance. The MSN model also achieves comparable acceleration ratios to the SOTA model with additional model structure on Spec-Bench.

Large Language Models (LLMs) demonstrate robust capabilities across various fields, leading to a paradigm shift in LLM-enhanced Recommender System (RS). Research to date focuses on point-wise and pair-wise recommendation paradigms, which are inefficient for LLM-based recommenders due to high computational costs. However, existing list-wise approaches also fall short in ranking tasks due to misalignment between ranking objectives and next-token prediction. Moreover, these LLM-based methods struggle to effectively address the order relation among candidates, particularly given the scale of ratings. To address these challenges, this paper introduces the large language model framework with Aligned Listwise Ranking Objectives (ALRO). ALRO is designed to bridge the gap between the capabilities of LLMs and the nuanced requirements of ranking tasks. Specifically, ALRO employs explicit feedback in a listwise manner by introducing soft lambda loss, a customized adaptation of lambda loss designed for optimizing order relations. This mechanism provides more accurate optimization goals, enhancing the ranking process. Additionally, ALRO incorporates a permutation-sensitive learning mechanism that addresses position bias, a prevalent issue in generative models, without imposing additional computational burdens during inference. Our evaluative studies reveal that ALRO outperforms both existing embedding-based recommendation methods and LLM-based recommendation baselines.

In the domain of Aspect-Based Sentiment Analysis (ABSA), generative methods have shown promising results and achieved substantial advancements. However, despite these advancements, the tasks of extracting sentiment quadruplets, which capture the nuanced sentiment expressions within a sentence, remain significant challenges. In particular, compound sentences can potentially contain multiple quadruplets, making the extraction task increasingly difficult as sentence complexity grows. To address this issue, we are focusing on simplifying sentence structures to facilitate the easier recognition of these elements and crafting a model that integrates seamlessly with various ABSA tasks. In this paper, we propose Aspect Term Oriented Sentence Splitter (ATOSS), which simplifies compound sentence into simpler and clearer forms, thereby clarifying their structure and intent. As a plug-and-play module, this approach retains the parameters of the ABSA model while making it easier to identify essential intent within input sentences. Extensive experimental results show that utilizing ATOSS outperforms existing methods in both ASQP and ACOS tasks, which are the primary tasks for extracting sentiment quadruplets

The limited generalization of coreference resolution (CR) models has been a major bottleneck in the task’s broad application. Prior work has identified annotation differences, especially for mention detection, as one of the main reasons for the generalization gap and proposed using additional annotated target domain data. Rather than relying on this additional annotation, we propose an alternative referential task, Major Entity Identification (MEI), where we: (a) assume the target entities to be specified in the input, and (b) limit the task to only the frequent entities. Through extensive experiments, we demonstrate that MEI models generalize well across domains on multiple datasets with supervised models and LLM-based few-shot prompting. Additionally, MEI fits the classification framework, which enables the use of robust and intuitive classification-based metrics. Finally, MEI is also of practical use as it allows a user to search for all mentions of a particular entity or a group of entities of interest.

Recent advancements in massively multilingual machine translation systems have significantly enhanced translation accuracy; however, even the best performing systems still generate hallucinations, severely impacting user trust. Detecting hallucinations in Machine Translation (MT) remains a critical challenge, particularly since existing methods excel with High-Resource Languages (HRLs) but exhibit substantial limitations when applied to Low-Resource Languages (LRLs). This paper evaluates sentence-level hallucination detection approaches using Large Language Models (LLMs) and semantic similarity within massively multilingual embeddings. Our study spans 16 language directions, covering HRLs, LRLs, with diverse scripts. We find that the choice of model is essential for performance. On average, for HRLs, Llama3-70B outperforms the previous state of the art by as much as 0.16 MCC (Matthews Correlation Coefficient). However, for LRLs we observe that Claude Sonnet outperforms other LLMs on average by 0.03 MCC. The key takeaway from our study is that LLMs can achieve performance comparable or even better than previously proposed models, despite not being explicitly trained for any machine translation task. However, their advantage is less significant for LRLs.

Referring Expression Comprehension (REC), which aims to ground a local visual region via natural language, is a task that heavily relies on multimodal alignment. Most existing methods utilize powerful pre-trained models to transfer visual/linguistic knowledge by full fine-tuning. However, full fine-tuning the entire backbone not only breaks the rich prior knowledge embedded in the pre-training, but also incurs significant computational costs. Motivated by the recent emergence of Parameter-Efficient Transfer Learning (PETL) methods, we aim to solve the REC task in an effective and efficient manner. Directly applying these PETL methods to the REC task is inappropriate, as they lack the specific-domain abilities for precise local visual perception and visual-language alignment. Therefore, we propose a novel framework of Multimodal Prior-guided Parameter Efficient Tuning, namely MaPPER. Specifically, MaPPER comprises Dynamic Prior Adapters guided by a aligned prior, and Local Convolution Adapters to extract precise local semantics for better visual perception. Moreover, the Prior-Guided Text module is proposed to further utilize the prior for facilitating the cross-modal alignment. Experimental results on three widely-used benchmarks demonstrate that MaPPER achieves the best accuracy compared to the full fine-tuning and other PETL methods with only 1.41% tunable backbone parameters.

Humans perform visual perception at multiple levels, including low-level object recognition and high-level semantic interpretation such as behavior understanding. Subtle differences in low-level details can lead to substantial changes in high-level perception. For example, substituting the shopping bag held by a person with a gun suggests violent behavior, implying criminal or violent activity. Despite significant advancements in various multimodal tasks, Large Visual Language Models (LVLMs) remain unexplored in their capabilities to conduct such multi-level visual perceptions.To investigate the perception gap between LVLMs and humans, we introduce MVP-Bench, the first visual–language benchmark systematically evaluating both low- and high-level visual perception of LVLMs. We construct MVP-Bench across natural and synthetic images to investigate how manipulated content influences model perception. Using MVP-Bench, we diagnose the visual perception of 10 open-source and 2 closed-source LVLMs, showing that high-level perception tasks significantly challenge existing LVLMs. The state-of-the-art GPT-4o only achieves an accuracy of 56% on Yes/No questions, compared with 74% in low-level scenarios. Furthermore, the performance gap between natural and manipulated images indicates that current LVLMs do not generalize in understanding the visual semantics of synthetic images as humans do.

Large Language Models (LLMs) are regularly updated to enhance performance, typically through changes in data or architecture. Within the update process, developers often prioritize improving overall performance metrics, paying less attention to maintaining compatibility with earlier model versions. Instance-level degradation (instance regression) of performance from one model version to the next can interfere with a user’s mental model of the capabilities of a particular language model. Users having to adapt their mental model with every update can lead to dissatisfaction, especially when the new model has degraded compared to a prior version for a known use case (model update regression).We find that when pretrained LLM base models are updated, fine-tuned user-facing downstream task adapters experience negative flips – previously correct instances are now predicted incorrectly. We observe model update regression between different model versions on a diverse set of tasks and models, even when the downstream task training procedures remain identical. We argue for the importance of maintaining model update compatibility during updates, and present evaluation metrics designed specifically for generative tasks, while also being applicable to discriminative tasks. We propose a training strategy to minimize the extent of instance regression in model updates, involving training of a compatibility adapter that can enhance task fine-tuned language models. We show negative flips reduce by up to 40% e.g. when updating Llama 1 to Llama 2 with our proposed method.

In linguistics, all languages can be considered as symbolic systems, with each language relying on symbolic processes to associate specific symbols with meanings. In the same language, there is a fixed correspondence between linguistic symbol and meaning. In different languages, universal meanings follow varying rules of symbolization in one-to-one correspondence with symbols. Most work overlooks the properties of languages as symbol systems. In this paper, we shift the focus to the symbolic properties and introduce MTLS: a pre-training method to improve the multilingual capability of models by Making Texts into Linguistic Symbols. Initially, we replace the vocabulary in pre-trained language models by mapping relations between linguistic symbols and semantics. Subsequently, universal semantics within the symbolic system serve as bridges, linking symbols from different languages to the embedding space of the model, thereby enabling the model to process linguistic symbols. To evaluate the effectiveness of MTLS, we conducted experiments on multilingual tasks using BERT and RoBERTa, respectively, as the backbone. The results indicate that despite having just over 12,000 pieces of English data in pre-training, the improvement that MTLS brings to multilingual capabilities is remarkably significant.

Although Dense Passage Retrieval (DPR) models have achieved significantly enhanced performance, their widespread application is still hindered by the demanding inference efficiency and high deployment costs. Knowledge distillation is an efficient method to compress models, which transfers knowledge from strong teacher models to weak student models. Previous studies have proved the effectiveness of knowledge distillation in DPR. However, there often remains a significant performance gap between the teacher and the distilled student. To narrow this performance gap, we propose MTA4DPR, a Multi-Teaching-Assistants based iterative knowledge distillation method for Dense Passage Retrieval, which transfers knowledge from the teacher to the student with the help of multiple assistants in an iterative manner; with each iteration, the student learns from more performant assistants and more difficult data. The experimental results show that our 66M student model achieves the state-of-the-art performance among models with same parameters on multiple datasets, and is very competitive when compared with larger, even LLM-based, DPR models.

Large language models (LLMs) are increasingly used for complex multi-turn conversations across diverse real-world applications. However, existing benchmarks mainly focus on single-turn evaluations, overlooking the models’ capabilities in multi-turn interactions. To address this gap, we introduce , a comprehensive benchmark to evaluate the multi-turn conversational abilities of LLMs. By analyzing human-LLM conversations, we categorize interaction patterns into four types: recollection, expansion, refinement, and follow-up. We construct multi-turn queries for each category either by augmenting existing datasets or creating new examples using GPT-4 with a human-in-the-loop process to avoid data leakage. To study the factors impacting multi-turn abilities, we create single-turn versions of the 1170 multi-turn queries and compare performance. Our evaluation of 10 well-known LLMs shows that while closed-source models generally surpass open-source ones, certain open-source models exceed GPT-3.5-Turbo in specific tasks. We observe significant performance degradation in multi-turn settings compared to single-turn settings in most models, which is not correlated with the models’ fundamental capabilities. Moreover, we identify the distance to relevant content and susceptibility to error propagation as the key factors influencing multi-turn performance.

Insight gradually becomes a crucial form of long-term memory for an agent. However, the emergence of irrelevant insight and the lack of general insight can greatly undermine the effectiveness of insight. To solve this problem, in this paper, we introduce **M**ulti-**S**cale **I**nsight Agent (MSI-Agent), an embodied agent designed to improve LLMs’ planning and decision-making ability by summarizing and utilizing insight effectively across different scales. MSI achieves this through the experience selector, insight generator, and insight selector. Leveraging a three-part pipeline, MSI can generate task-specific and high-level insight, store it in a database, and then use relevant insight from it to aid in decision-making. Our experiments show that MSI outperforms another insight strategy when planning by GPT3.5. Moreover, We delve into the strategies for selecting seed experience and insight, aiming to provide LLM with more useful and relevant insight for better decision-making. Our observations also indicate that MSI exhibits better robustness when facing domain-shifting scenarios.

Knowledge graph embedding (KGE) models achieved state-of-the-art results on many knowledge graph tasks including link prediction and information retrieval. Despite the superior performance of KGE models in practice, we discover a deficiency in the expressiveness of some popular existing KGE models called Z-paradox. Motivated by the existence of Z-paradox, we propose a new KGE model called MQuinE that does not suffer from Z-paradox while preserves strong expressiveness to model various relation patterns including symmetric/asymmetric, inverse, 1-N/N-1/N-N, and composition relations with theoretical justification. Experiments on real-world knowledge bases indicate that Z-paradox indeed degrades the performance of existing KGE models, and can cause more than 20% accuracy drop on some challenging test samples. Our experiments further demonstrate that MQuinE can mitigate the negative impact of Z-paradox and outperform existing KGE models by a visible margin on link prediction tasks.

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

RNA foundation models (FMs) have been extensively used to interpret genomic sequences and address a wide range of in-silico genomic tasks. However, current RNA FMs often overlook the incorporation of secondary structures in the pretraining of FMs, which impedes the effectiveness in various genomic tasks. To address this problem, we leverage filtered high-fidelity structure annotations for structure pretraining to enhance the modeling ability of FMs in single nucleotide resolution tasks. Experimental evaluations across four comprehensive genomic benchmarks demonstrate that our RNA FM consistently outperforms existing RNA FMs, achieving a 40% improvement in RNA secondary structure prediction and obtaining top-tier results on DNA genomic benchmarks even though it has not been pretrained on any DNA genome. We release the code and models to encourage further research to bridge the gap between in-silico predictions and biological reality.

Rapid advancements over the years have helped machine learning models reach previously hard-to-achieve goals, sometimes even exceeding human capabilities. However, achieving desired accuracy comes at the cost of larger model sizes and increased computational demands. Thus, serving predictions from these models to meet any latency and cost requirements of applications remains a key challenge, despite recent work in building inference serving systems as well as algorithmic approaches that dynamically adapt models based on inputs. Our paper introduces a new form of dynamism, modality selection, where we adaptively choose modalities from inference inputs while maintaining the model quality. We introduce MOSEL, an automated inference serving system for multi-modal ML models that carefully picks input modalities per request based on user-defined performance and accuracy requirements. MOSEL exploits modality configurations extensively, improving system throughput by 3.6 × with an accuracy guarantee. It also reduces job completion times by 11× compared to modality-agnostic approaches.

The rise of foundation models (FMs), coupled with regulatory efforts addressing their risks and impacts, has sparked significant interest in open-source models. However, existing speech FMs (SFMs) fall short of full compliance with the open-source principles, even if claimed otherwise, as no existing SFM has model weights, code, and training data publicly available under open-source terms. In this work, we take the first step toward filling this gap by focusing on the 24 official languages of the European Union (EU). We collect suitable training data by surveying automatic speech recognition datasets and unlabeled speech corpora under open-source compliant licenses, for a total of 950k hours. Additionally, we release automatic transcripts for 441k hours of unlabeled data under the permissive CC-BY license, thereby facilitating the creation of open-source SFMs for the EU languages.

Personalized Dialogue Generation (PDG) aims to create coherent responses according to roles or personas. Traditional PDG relies on external role data, which can be scarce and raise privacy concerns. Approaches address these issues by extracting role information from dialogue history, which often fail to generically model roles in continuous space. To overcome these limitations, we introduce a novel framework Models Roles from Personalized Dialogue History by Exploring and Utilizing Latent Space (MORPHEUS) through a three-stage training process. Specifically, we create a persona codebook to represent roles in latent space compactly, and this codebook is used to construct a posterior distribution of role information. This method enables the model to generalize across roles, allowing the generation of personalized dialogues even for unseen roles. Experiments on both Chinese and English datasets demonstrate that MORPHEUS enhances the extraction of role information, and improves response generation without external role data. Additionally, MORPHEUS can be considered an efficient fine-tuning for large language models.