Austria

We introduce \textbf{Dynamic Retrieval-Augmented Expert Networks (DRAE)}, a groundbreaking architecture that addresses the challenges of lifelong learning, catastrophic forgetting, and task adaptation by combining the dynamic routing capabilities of Mixture-of-Experts (MoE); leveraging the knowledge-enhancement power of Retrieval-Augmented Generation (RAG); incorporating a novel hierarchical reinforcement learning (RL) framework; and coordinating through ReflexNet-SchemaPlanner-HyperOptima (RSHO).DRAE dynamically routes expert models via a sparse MoE gating mechanism, enabling efficient resource allocation while leveraging external knowledge through parametric retrieval (P-RAG) to augment the learning process. We propose a new RL framework with ReflexNet for low-level task execution, SchemaPlanner for symbolic reasoning, and HyperOptima for long-term context modeling, ensuring continuous adaptation and memory retention. Experimental results show that DRAE significantly outperforms baseline approaches in long-term task retention and knowledge reuse, achieving an average task success rate of 82.5\% across a set of dynamic robotic manipulation tasks, compared to 74.2\% for traditional MoE models. Furthermore, DRAE maintains an extremely low forgetting rate, outperforming state-of-the-art methods in catastrophic forgetting mitigation. These results demonstrate the effectiveness of our approach in enabling flexible, scalable, and efficient lifelong learning for robotics.

ACL 2025

DRAE: Dynamic Retrieval-Augmented Expert Networks for Lifelong Learning and Task Adaptation in Robotics

retrieval-augmented generation

catastrophic forgetting

lifelong learning

## 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*

EMNLP 2025

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

technical paper

### Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Message from the General Chair: 
*It is my great pleasure and honor to welcome you to the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025), held in beautiful Vienna, Austria, from July 27 to August 1, 2025. ACL2025continues our field’s tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.* 
[Read more](https://drive.google.com/file/d/1GI_hvOpjswAuYdUTromfeDiPpCcqidwg/view?usp=sharing)

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2025". Please be sure to check your spam email folder if you do not see an email confirmation right away.

Please log in

To access this event page, you need to log in with the **email address you registered with**. 

Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Recently, a number of repository-level code generation benchmarks--such as CoderEval, DevEval, RepoEval, RepoBench, and LongCode-Arena--have emerged to evaluate the capabilities of large language models (LLMs) beyond standalone benchmarks like HumanEval and MBPP. Thus, a natural question is, would LLMs have similar performance in real world coding tasks as their performance in these benchmarks? Unfortunately, one cannot answer this question, since these benchmarks consist of short completions, synthetic examples, or focus on limited scale repositories, failing to represent real-world coding tasks.

To address these challenges, we create RepoCod, a Python code-generation benchmark containing complex tasks with realistic dependencies in real-world large projects and appropriate metrics for evaluating source code. It includes 980 whole-function generation tasks from 11 popular projects, 50.8\% of which require repository-level context. RepoCod includes 314 developer-written test cases per instance for better evaluation. We evaluate ten LLMs on RepoCod and find that none achieves more than 30\% pass@1 on RepoCod, indicating the necessity of building stronger LLMs that can help developers in real-world software development. In addition, we found that retrieval-augmented generation achieves better results than using target function dependencies as context.

Can Language Models Replace Programmers? REPOCOD Says ‘Not Yet’

Web development involves turning UI designs into functional webpages, which can be difficult for both beginners and experienced developers due to the complexity of HTML's hierarchical structures and styles. While Large Language Models (LLMs) have shown promise in generating source code, two major challenges persist in UI-to-HTML code generation: (1) effectively representing HTML's hierarchical structure for LLMs, and (2) bridging the gap between the visual nature of UI designs and the text-based format of HTML code. To tackle these challenges, we introduce Waffle, a new fine-tuning strategy that uses a structure-aware attention mechanism to improve LLMs' understanding of HTML's structure and a contrastive fine-tuning approach to align LLMs' understanding of UI images and HTML code. Models fine-tuned with Waffle show up to 9.00 pp (percentage point) higher HTML match, 0.0982 higher CW-SSIM, 32.99 higher CLIP, and 27.12 pp higher LLEM on our new benchmark WebSight-Test and an existing benchmark Design2Code, outperforming current fine-tuning methods.

WAFFLE: Fine-tuning Multi-Modal Model for Automated Front-End Development

Multimodal summarization (MS) combines text and visuals to generate summaries. Recently, many-to-many multimodal summarization (M3S) garnered interest as it enables a unified model for multilingual and cross-lingual MS. Existing methods have made progress by facilitating the transfer of common multimodal summarization knowledge. While, prior M3S models that fully share parameters neglect the language-specific knowledge learning, where potential interference between languages may limit the flexible adaptation of MS modes across different language combinations and hinder further collaborative improvements in joint M3S training. Based on this observation, we propose Language Constrained Multimodal Hyper Adapter (LCMHA) for M3S. LCMHA integrates language-specific multimodal adapters into multilingual pre-trained backbones via a language constrained hypernetwork, enabling relaxed parameter sharing that enhances language-specific learning while preserving shared MS knowledge learning. In addition, a language-regularized hypernetwork is designed to balance intra- and inter-language learning, generating language-specific adaptation weights and enhancing the retention of distinct language features through the regularization of generated parameters. Experimental results on the M3Sum benchmark show LCMHA's effectiveness and scalability across multiple multilingual pre-trained backbones.

Language Constrained Multimodal Hyper Adapter For Many-to-Many Multimodal Summarization

LLMs have improved the fluency and informativeness of abstractive summarization but remain prone to hallucinations, where generated content deviates from the source document. Recent PMI decoding strategies mitigate over-reliance on prior knowledge by comparing output probabilities with and without source documents, effectively enhancing contextual utilization and improving faithfulness. However, existing strategies often neglect the explicit use of salient contextual information and rely on static hyperparameters to fix the balance between contextual and prior knowledge, limiting their flexibility. In this work, we propose Salience-Aware Reinforced Adaptive decoding (SARA), which incorporates salient information and allows the model to adaptively determine reliance on the source document's context, salient context, and the model's prior knowledge based on pointwise mutual information. Moreover, a tokenwise adaptive decoding mechanism via reinforcement learning is proposed in SARA to dynamically adjust the contributions of context and prior knowledge at each decoding timestep. Experiments on CNN/DM, WikiHow, and NYT50 datasets show that SARA consistently improves the quality and faithfulness of summaries across various LLM backbones without modifying their weights.

SARA: Salience-Aware Reinforced Adaptive Decoding for Large Language Models in Abstractive Summarization

Auditory foundation models, including auditory large language models (LLMs), process all sound inputs equally, independent of listener perception. However, human auditory perception is inherently selective: listeners focus on specific speakers while ignoring others in complex auditory scenes. Existing models do not incorporate this selectivity, limiting their ability to generate perception-aligned responses. To address this, we introduce intention-informed auditory scene understanding (II-ASU) and present Auditory Attention-Driven LLM (AAD-LLM), a prototype system that integrates brain signals to infer listener attention. AAD-LLM extends an auditory LLM by incorporating intracranial electroencephalography (iEEG) recordings to decode which speaker a listener is attending to and refine responses accordingly. The model first predicts the attended speaker from neural activity, then conditions response generation on this inferred attentional state. We evaluate AAD-LLM on speaker description, speech transcription and extraction, and question answering in multitalker scenarios, with both objective and subjective ratings showing improved alignment with listener intention. By taking a first step toward intention-aware auditory AI, this work explores a new paradigm where listener perception informs machine listening, paving the way for future listener-centered auditory systems. Demo available.

AAD-LLM: Neural Attention-Driven Auditory Scene Understanding

The few-shot relation extraction (FSRE) aims at enhancing the model’s generalization to new relations with very few labeled instances (support instances). Most existing studies use prototype networks (ProtoNets) for FSRE and assume that the support set, adapting the model to new relations, only contains accurately labeled instances. However, this assumption is usually unrealistic, as even carefully-annotated datasets often contain mislabeled instances. Thus, it is essential to enhance the robustness of FSRE models to noisy labels in support set, but this issue remains unexplored. In this paper, we first conduct a preliminary study, revealing the high sensitivity of ProtoNets to such noisy labels. Meanwhile, we discover that fully leveraging mislabeled support instances is crucial for enhancing the model's robustness. To do this, we propose a self-denoising model for FSRE, which can automatically correct noisy labels of support instances. Specifically, our model comprises two core components: 1) a label correction module (LCM), used to correct mislabeled support instances based on the distances between them in the embedding space, and 2) a relation classification module (RCM), designed to achieve more robust relation prediction using the corrected labels generated by the LCM. Moreover, we propose a feedback-based training strategy, which focuses on training LCM and RCM to synergistically handle noisy labels in support set. Experimental results on two public datasets show the effectiveness and robustness of our model. Notably, even in scenarios without noisy labels, our model significantly outperforms all competitive baselines.

A Self-Denoising Model for Robust Few-Shot Relation Extraction

Automated radiology report generation from chest X-ray (CXR) images has the potential to improve clinical efficiency and reduce radiologists' workload. However, most datasets, including the publicly available MIMIC-CXR and CheXpert Plus, consist entirely of free-form reports, which are inherently variable and unstructured. This variability poses challenges for both generation and evaluation: existing models struggle to produce consistent, clinically meaningful reports, and standard evaluation metrics fail to capture the nuances of radiological interpretation. To address this, we introduce Structured Radiology Report Generation (SRRG), a new task that reformulates free-text radiology reports into a standardized format, ensuring clarity, consistency, and structured clinical reporting. We create a novel dataset by restructuring reports using large language models (LLMs) following strict structured reporting desiderata. Additionally, we introduce SRR-BERT, a fine-grained disease classification model trained on 55 labels, enabling more precise and clinically informed evaluation of structured reports. To assess report quality, we propose F1-SRR-BERT, a metric that leverages SRR-BERT’s hierarchical disease taxonomy to bridge the gap between free-text variability and structured clinical reporting. We validate our dataset through a reader study conducted by five board-certified radiologists and extensive benchmarking experiments.

Automated Structured Radiology Report Generation

This paper argues that generating output tokens is more effective than using pooled representations for prediction tasks because token-level generation retains more mutual information. Since LLMs are trained on massive text corpora using next-token prediction, generation aligns naturally with their learned behavior. Using the Data Processing Inequality (DPI), we provide both theoretical and empirical evidence supporting this claim. However, autoregressive models face two key challenges when used for prediction: (1) exposure bias, where the model sees ground-truth tokens during training but relies on its own predictions during inference, leading to errors, and (2) format mismatch, where discrete tokens do not always align with the task’s required output structure. To address these challenges, we introduce PredGen (Predicting Through Generating), an end-to-end framework that (i) uses scheduled sampling to reduce exposure bias, and (ii) introduces a task adapter to convert the generated tokens into structured outputs. Additionally, we introduce Writer-Director Alignment Loss (WDAL), which ensures consistency between token generation and final task predictions, improving both text coherence and numerical accuracy. We evaluate PredGen on multiple classification and regression benchmarks. Our results show that PredGen consistently outperforms standard baselines, demonstrating its effectiveness in structured prediction tasks.

Predicting Through Generation: Why Generation Is Better for Prediction

Multi-modal entity alignment aims to identify equivalent entities between two different multi-modal knowledge graphs. Current methods have made significant progress by improving embedding and cross-modal fusion. However, most of them depend on using loss functions to capture the relationship between modalities or adopt a one-time strategy to directly compute modality weights using attention mechanisms, which overlooks the relative interactions between modalities at the entity level and the accuracy of modality weights, thereby hindering the generalization to diverse entities. To address this challenge, we propose RICEA, a relative interaction and calibration framework for multi-modal entity alignment, which dynamically computes weights based on the relative interaction and recalibrates the weights according to their uncertainties. Among these, we propose a novel method called ADC that utilizes attention mechanisms to perceive the uncertainty of the weight for each modality, rather than directly calculating the weight of each modality as in previous works. Across 5 datasets and 23 settings, our proposed framework significantly outperforms other baselines. Our code and data are available at https://github.com/ChenxiaoLi-Joe/RICEA.

Probing Relative Interaction and Dynamic Calibration in Multi-modal Entity Alignment

While Large Language Models (LLMs) excel in diverse domains, their validity in event reasoning remains underexplored. Most existing works merely stagnate at assessing LLMs' event reasoning with a single event relational type or reasoning format, failing to conduct a complete evaluation and provide a practical solution for capability enhancement. In this paper, we propose $\textbf{PIPER}$, the first comprehensive benchmark for $\textbf{P}$robing $\textbf{I}$nto the $\textbf{P}$erformance boundary of LLMs in $\textbf{E}$vent $\textbf{R}$easoning. Motivated by our evaluation observations and error patterns analysis, we meticulously craft 10K diverse instruction-tuning demonstrations to alleviate event reasoning-oriented data scarcity. Additionally, a novel $\textbf{D}$ebiasing and $\textbf{D}$istillation-$\textbf{E}$nhanced $\textbf{S}$upervised $\textbf{F}$ine-$\textbf{T}$uning ($\mathbf{D^2}$$\textbf{E-SFT}$) strategy is presented, which facilitates adhering to context and fixating significant contextual event information to elevate the event reasoning capability. Specifically, $\mathrm{D^2}$E-SFT removes the given sample's context to construct an imagined sample, subtracting its logits to mitigate the bias of neglecting context and improve contextual faithfulness. To guide the model in emphasizing significant contextual event information, $\mathrm{D^2}$E-SFT employs a context-refined sample to achieve self-distillation with the alignment of logits. Extensive experimental results demonstrate the effectiveness of our data and strategy in expanding the performance boundary of event reasoning.

Downloads

Next from ACL 2025

Can Language Models Replace Programmers? REPOCOD Says ‘Not Yet’

Similar lecture

Dynamic Expert Specialization: Towards Catastrophic Forgetting-Free Multi-Domain MoE Adaptation