China

While training large language models (LLMs) from scratch can indeed lead to models with distinct capabilities and strengths, it incurs substantial costs and may lead to redundancy in competencies. Knowledge fusion aims to integrate existing LLMs of diverse architectures and capabilities into a more potent LLM through lightweight continual training, thereby reducing the need for costly LLM development. In this work, we propose a new framework for the knowledge fusion of chat LLMs through two main stages, resulting in FuseChat. Firstly, we conduct pairwise knowledge fusion on source chat LLMs of varying structures and scales to create multiple target LLMs with identical structure and size via lightweight fine-tuning. During this process, a statistics-based token alignment approach is introduced as the cornerstone for fusing LLMs with different structures. Secondly, we merge these target LLMs within the parameter space, where we propose a novel method for determining the merging coefficients based on the magnitude of parameter updates before and after fine-tuning. We implement and validate FuseChat using six prominent chat LLMs with diverse architectures and scales. Experimental results on two instruction-following benchmarks, AlpacaEval 2.0 and MT-Bench, demonstrate the superiority of FuseChat-7B over baselines of various sizes.

EMNLP 2025

FuseChat: Knowledge Fusion of Chat Models

model fusion

alignment

large language models

### Welcome!
The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. Our Virtual Poster Sessions will take place online Thursday, August 22, 2024.

You are required to register for this event. **Please register [here](https://2024.aclweb.org/registration). **

If you have already registered, please check your inbox for an email from Underline granting you access to ACL 2024 content.

Please register!

ACL 2024

The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. More information will be announced soon.

poster

## Welcome!
"I am excited to welcome you to this year’s edition of the Conference on Empirical Methods in Natural Language Processing! Importantly, it marks the 30th edition of EMNLP. With over 8,000 submissions, more than 3,000 accepted papers, and thousands of attendees, we have come a long way from that first
workshop, which had 14 accepted papers. As the field looks ahead, Suzhou is the fitting location for celebrating this milestone: rooted in a long literary tradition, yet modern and forward-looking, and home to a large share of the NLP community." 

*Message from the General Chair, Dirk Hovy*

Celebrate 30 Years of EMNLP! 
EMNLP 2025 will be held in Suzhou, China from November 5th to November 9th, 2025.

End-to-end automatic speech recognition (ASR) based on deep learning has achieved impressive progress in recent years. However, the performance of ASR foundation model often degrades significantly on out-of-domain data due to real-world domain shifts. Test-Time Adaptation (TTA) methods aim to mitigate this issue by adapting models during inference without access to source data. Despite recent progress, existing ASR TTA methods often struggle with instability under continual and long-term distribution shifts. To alleviate the risk of performance collapse due to error accumulation, we propose Dynamic Model-bank Single-Utterance Test-time Adaptation (DMSUTA), a sustainable continual TTA framework based on adaptive ASR model ensembling. DMSUTA maintains a dynamic model bank, from which a subset of checkpoints is selected for each test sample based on confidence and uncertainty criteria. To preserve both model plasticity and long-term stability, DMSUTA actively manages the bank by filtering out potentially collapsed models. This design allows DMSUTA to continually adapt to evolving domain shifts in ASR test-time scenarios. Experiments on diverse, continuously shifting ASR TTA benchmarks show that DMSUTA consistently outperforms existing continual TTA baselines, demonstrating superior robustness to domain shifts in ASR.

Dynamic Model-Bank Test-Time Adaptation for Automatic Speech Recognition

Goal-oriented dialogues, such as recommendation and negotiation, often require balancing multiple, conflicting objectives. Existing methods typically involve training separate models for specific combinations of objectives, leading to computational and scalability issues. In this work, we aim to develop a new dialogue policy method that can adapt to varying objective preferences at inference time without retraining. This raises several challenges in terms of both (1) optimization strategy and (2) knowledge utilization. To address these, we propose a novel learning framework, Preference Adaptive Dialogue Policy Planner (PADPP), for multi-objective goal-oriented dialogues. Specifically, to tackle the former, we introduce a novel policy optimization scheme, which leverages information gained from training the model on previously updated objective weights, accelerating the learning capability on new weight settings. To address the latter, we utilize Generalized Policy Improvement (GPI) to ensure the effectiveness of leveraged knowledge. Experimental results demonstrate that PADPP achieves superior adaptability and performance compared to state-of-the-art approaches, offering a scalable and flexible solution for multi-objective, goal-oriented dialogues. Code and data are available at the anonymous link.

One Planner To Guide Them All ! Learning Adaptive Conversational Planners for Goal-oriented Dialogues

Text embeddings and text embedding models are bread-and-butter in many NLP tasks, especially in those that involve classification, clustering or search. However, interpretability challenges persist, especially in explaining obtained similarity scores, which is crucial for applications requiring transparency. In this piece, we give a structured overview of methods specializing in explaining those similarity scores, an interesting and fairly unexplored research area. Specifically, we highlight the methods' individual ideas and techniques, as well as their trade-offs, assessing their potential for interpreting text embeddings and explaining similarity. We finally outline opportunities and open challenges for future research.

Interpretable Text Embeddings and Text Similarity Explanation: A Survey

Detecting offensive language in Chinese is challenging due to homophonic substitutions used to evade detection. We propose a framework to improve large language models’ robustness against such phonetic attacks. First, we construct HED-COLD, a homophone-enhanced dataset based on the Chinese Offensive Language Dataset. Additionally, we propose a homophone-aware pretraining strategy that aligns semantics and fuses features to learn robust mappings between original and perturbed text. Experimental results show that our approach achieves state-of-the-art performance on both the COLD test set and the toxicity benchmark ToxiCloakCN. Notably, it achieves greater gains in domains especially prone to homophonic attacks, such as gender and regional content. These results demonstrate improved robustness and generalization against phonetic adversarial attacks.

Enhancing Chinese Offensive Language Detection with Homophonic Perturbation

Disentangling how gender and occupations are encoded by LLMs is crucial to identify possible biases and prevent harms, especially given the widespread use of LLMs in sensitive domains such as human resources. In this work, we carry out an in-depth investigation of gender and occupational biases in English and Italian as expressed by 9 different LLMs (both base and instruction-tuned). Specifically, we focus on the analysis of sentence completions when LLMs are prompted with job-related sentences including different gender representations. We carry out a manual analysis of 4,500 generated texts over 4 dimensions that can reflect bias, we propose a novel embedding-based method to investigate biases in generated texts and, finally, we carry out a lexical analysis of the model completions. In our qualitative and quantitative evaluation we show that many facets of social bias remain unaccounted for even in aligned models, and LLMs in general still reflect existing gender biases in both languages. Finally, we find that models still struggle with gender-neutral expressions, especially beyond English.

Job Unfair: An Investigation of Gender and Occupational Bias in Free-Form Text Completions by LLMs

Recently, Spoken Dialogue Models (SDMs) have attracted significant attention for their ability to generate voice responses directly to users' spoken queries. Despite their increasing popularity, there exists a gap in research focused on comprehensively understanding their practical effectiveness in comprehending and emulating human conversations. This is especially true when compared to text-based Large Language Models (LLMs), which benefit from extensive benchmarking. Human voice interactions are inherently more complex than text due to characteristics unique to spoken dialogue. Ambiguity poses one challenge, stemming from semantic factors like polysemy, as well as phonological aspects such as heterograph, heteronym, and stress patterns. Additionally, context-dependency like omission, coreference and multi-turn interaction adds further complexity to human conversational dynamics. To illuminate the current state of SDM development and to address these challenges, we present a benchmark dataset in this paper, which comprises 1,079 samples in English and Chinese. Accompanied by an LLM-based evaluation method that closely aligns with human judgment, this dataset facilitates a comprehensive exploration of the performance of SDMs in tackling these practical challenges.

C3: A Bilingual Benchmark for Spoken Dialogue Models Exploring Challenges in Complex Conversations

Large language models offer transformative potential for healthcare, yet their responsible and equitable development depends critically on a deeper understanding of how training data characteristics influence model behavior, including the potential for bias. Current practices in dataset curation and bias assessment often lack the necessary transparency, creating an urgent need for comprehensive evaluation frameworks to foster trust and guide improvements. In this study, we present an in-depth analysis of potential downstream biases in clinical language models, with a focus on differential opioid prescription tendencies across diverse demographic groups, such as ethnicity, gender, and age. As part of this investigation, we introduce HC4: Healthcare Comprehensive Commons Corpus, a novel and extensively curated pretraining dataset exceeding 89 billion tokens. Our evaluation leverages both established general benchmarks and a novel, healthcare-specific methodology, offering crucial insights to support fairness and safety in clinical AI applications.

Building Trust in Clinical LLMs: Bias Analysis and Dataset Transparency

The formation and circulation of ideas in philosophy have profound implications for understanding philosophical dynamism--enabling us to identify seminal texts, delineate intellectual traditions, and track changing conventions in the act of philosophizing. However, traditional analyses of these issues often depend on manual reading and subjective interpretation, constrained by human cognitive limits. We introduce InterIDEAS, a pioneering dataset designed to bridge philosophy, literary studies, and natural language processing (NLP). By merging theories of intertextuality from literary studies with bibliometric techniques and recent LLMs, InterIDEAS enables both quantitative and qualitative analysis of the intellectual, social, and historical relations embedded within authentic philosophical texts. This dataset not only assists the study of philosophy but also contributes to the development of language models by providing a training corpus that challenges and enhances their interpretative capacity.

InterIDEAS: Philosophical Intertextuality via LLMs

Different languages are known to have typical and distinctive prosodic profiles. However, the majority of work on prosody across languages has been restricted to monolingual discourse contexts. We build on prior studies by asking: how does the nature of the discourse context influence variations in the prosody of monolingual speech? To answer this question, we compare the prosody of spontaneous and conversational monolingual English and Spanish both in monolingual \emph{and} in multilingual speech settings. For both languages, we find that monolingual speech produced in a \emph{monolingual} context is prosodically different from that produced in a \emph{multilingual} context, with more marked differences having increased proximity to multilingual discourse. Our work is the first to incorporate multilingual discourse contexts into the study of native-level monolingual prosody, and has potential downstream applications for the recognition and synthesis of multilingual speech.

Does Context Matter? A Prosodic Comparison of English and Spanish in Monolingual and Multilingual Discourse Settings

Current speech translation systems, while having achieved impressive accuracies, are rather static in their behavior and do not adapt to real-world situations in ways human interpreters do. In order to improve their practical usefulness and enable interpreting-like experiences, a precise understanding of the nature of human interpreting is crucial. To this end, we discuss human interpreting literature from the perspective of the machine translation field, while considering both operational and qualitative aspects. We identify implications for the development of speech translation systems and argue that there is great potential to adopt many human interpreting principles using recent modeling techniques. We hope that our findings provide inspiration for closing the perceived usability gap, and can motivate progress toward true machine interpreting.

Downloads

Next from EMNLP 2025

Dynamic Model-Bank Test-Time Adaptation for Automatic Speech Recognition

Similar lecture

Chat Vector: A Simple Approach to Equip LLMs with Instruction Following and Model Alignment in New Languages