Austria

While large language models (LLMs) have significantly advanced mathematical reasoning, Process Reward Models (PRMs) have been developed to evaluate the logical validity of reasoning steps. However, PRMs still struggle with out-of-distribution (OOD) challenges. This paper identifies the OOD issues including step OOD, arising from differences in reasoning patterns across model types and sizes, and question OOD, due to dataset shifts between training and real-world problems. To address these issues, we introduce Retrieval-Augmented Process Reward Model (RetrievalPRM), a novel framework designed to tackle these OOD issues. By utilizing a two-stage retrieval-enhanced mechanism, RetrievalPRM retrieves semantically similar questions and steps for PRM as a warmup to stimulate its potential to judge target steps, improving generalization and reasoning consistency across different models and problem types. Our extensive experiments demonstrate that RetrievalPRM outperforms existing baselines across multiple real-world datasets. Our open-source contributions include a retrieval-enhanced dataset, a tuning framework for PRM training, and the RetreivalPRM model, establishing a new standard for PRM performance.

ACL 2025

Retrieval-Augmented Process Reward Model for Generalizable Mathematical Reasoning

process reward model

retrieval-augmented generation

mathematical reasoning

## 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.

poster

### 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)

The latent knowledge of large language models (LLMs) contains harmful or unethical content, which introduces significant security risks upon their widespread deployment. Conducting jailbreak attacks on LLMs can proactively identify vulnerabilities to enhance their security measures. However, previous jailbreak attacks primarily focus on single-turn dialogue scenarios, leaving vulnerabilities in multi-turn dialogue contexts inadequately explored. This paper investigates the resilience of black-box LLMs in multi-turn jailbreak attack scenarios from a novel interrogation perspective. We propose an optimal interrogation principle to conceal the jailbreak intent and introduce a multi-turn attack chain generation strategy called CoA. By employing two effective interrogation strategies tailored for LLMs, coupled with an interrogation history record management mechanis, it achieves a significant optimization of the attack process. Our approach enables the iterative generation of attack chains, offering a powerful tool for LLM red team testing. Experimental results demonstrate that LLMs exhibit insufficient resistance under multi-turn interrogation, with our method shows more advantages(ASR, 83% vs 64%). This work offers new insights into improving the safety of LLMs.

Chain of Attack: Hide Your Intention through Multi-Turn Interrogation

The rapid adoption of large language models (LLMs) in diverse applications has intensified concerns over their security and integrity, especially in cloud environments where internal model parameters are inaccessible to users. Traditional tamper detection methods, designed for deterministic classification models, fail to address the output randomness and massive parameter spaces characteristic of LLMs. In this paper, we introduce \emph{Efficient Sensitive Fingerprinting (ESF)}, the first fingerprinting method tailored for black-box tamper detection of LLMs. ESF generates fingerprint samples by optimizing output sensitivity at selected detection token positions and leverages \emph{Randomness-Set Consistency Checking (RSCC)} to accommodate inherent output randomness. Furthermore, a novel \emph{Max Coverage Strategy (MCS)} is proposed to select an optimal set of fingerprint samples that maximizes joint sensitivity to tampering. Grounded in a rigorous theoretical framework, ESF is both computationally efficient and scalable to large models. Extensive experiments across state-of-the-art LLMs demonstrate that ESF reliably detects tampering, such as fine-tuning, model compression, and backdoor injection, with a detection rate exceeding 99.2\%, thereby offering a robust solution for securing cloud-based AI systems.

ESF: Efficient Sensitive Fingerprinting for Black-Box Tamper Detection of Large Language Models

Solving expert-level multimodal tasks is a key milestone in general intelligence. As the capabilities of multimodal large language models (MLLMs) continue to evolve, evaluation of frontier multimodal intelligence becomes necessary yet challenging. In this work, we introduce ProBench, a benchmark of open-ended user queries encapsulating professional expertise and advanced reasoning. ProBench consists of 4,000 high-quality samples independently collected from professionals based on their productivity demands. It spans across 10 fields and 56 sub-fields, including science, arts, humanities, coding, mathematics, and creative writing. Experimentally, we evaluate and compare 24 latest models using MLLM-as-a-Judge. Our results reveal that although the best open-source models rival the proprietary ones, they all face significant challenges in visual perception, textual understanding, domain knowledge, and advanced reasoning. Our benchmark is publicly accessible at \url{TBC}.

ProBench: Judging Multimodal Foundation Models on Open-ended Multi-domain Expert Tasks

Large language models (LLMs) exhibit remarkable multilingual capabilities despite the extreme language imbalance in the pre-training data. In this paper, we closely examine the reasons behind this phenomenon, focusing on the pre-training corpus. We find that the existence of code-switching, alternating between different languages within a context, is key to multilingual capabilities. We conduct an analysis to investigate code-switching in the pre-training corpus, examining its presence and categorizing it into four types within two quadrants. We then assess its impact on multilingual performance. These types of code-switching data are unbalanced in proportions and demonstrate different effects on facilitating language transfer. To better explore the power of code-switching for language alignment during pre-training, we investigate the strategy of synthetic code-switching. We continuously scale up the synthetic code-switching data and observe remarkable improvements in both benchmarks and representation space. Extensive experiments indicate that incorporating synthetic code-switching data enables better language alignment and generalizes well to high, medium, and low-resource languages with pre-training corpora of varying qualities.

Investigating and Scaling up Code-Switching for Multilingual Language Model Pre-Training

Sarcasm is a complex form of sentiment expression widely used in human daily life. Previous work primarily defines sarcasm as a form of verbal irony, which covers only a subset of real-world sarcastic expressions. However, sarcasm serves multifaceted functions and manifests itself through various rhetorical devices, such as echoic mention, rhetorical question and hyperbole. To fully capture its complexity, this paper investigates fine-grained sarcasm classification through the lens of rhetorical devices, and introduces RedSD, a RhEtorical Device-Aware Sarcasm Dataset with counterfactually augmented data.To construct the dataset, we extract sarcastic dialogues from situation comedies (i.e., sitcoms), and summarize nine rhetorical devices commonly employed in sarcasm. We then propose a rhetorical device-aware counterfactual data generation pipeline facilitated by both Large Language Models (LLMs) and human revision. Additionally, we propose duplex counterfactual augmentation that generates counterfactuals for both sarcastic and non-sarcastic dialogues, to further enhance the scale and diversity of the dataset.Experimental results on the dataset demonstrate that fine-tuned models exhibit a more balanced performance compared to zero-shot models, including GPT-3.5 and LLaMA 3.1, underscoring the importance of integrating various rhetorical devices in sarcasm detection. Our dataset is avaliable at https://github.com/qqHong73/RedSD.

Rhetorical Device-Aware Sarcasm Detection with Counterfactual Data Augmentation

Despite advancements in Natural Language Processing (NLP), models remain vulnerable to adversarial attacks, such as synonym substitutions. While prior work has focused on improving robustness for feed-forward and convolutional architectures, the robustness of recurrent networks and modern state space models (SSMs), such as S4, remains understudied. These architectures pose unique challenges due to their sequential processing and complex parameter dynamics. In this paper, we introduce a novel regularization technique based on Growth Bound Matrices (GBM) to improve NLP model robustness by reducing the impact of input perturbations on model outputs. We focus on computing the GBM for three architectures: Long Short-Term Memory (LSTM), State Space models (S4), and Convolutional Neural Networks (CNN). Our method aims to (1) enhance resilience against word substitution attacks, (2) improve generalization on clean text, and (3) providing the first systematic analysis of SSM (S4) robustness. Extensive experiments across multiple architectures and benchmark datasets demonstrate that our method improves adversarial robustness by up to (8.8%) over existing baselines. These results highlight the effectiveness of our approach, outperforming several state-of-the-art methods in adversarial defense.

Bridging Robustness and Generalization Against Word Substitution Attacks in NLP via the Growth Bound Matrix Approach

In-context learning (ICL) relies heavily on high-quality demonstrations drawn from large annotated corpora. Existing methods often detect noisy annotations by ranking local perplexities, presuming that noisy samples yield higher perplexities than their clean counterparts. However, this assumption breaks down when the noise ratio is high and many demonstrations are flawed. We re-examine the perplexity-based paradigm for text generation under noisy annotations, highlighting two sources of bias in perplexity: the annotation itself and the domain-specific knowledge inherent in large language models (LLMs). To overcome these biases, we introduce a dual-debiasing framework that uses synthesized neighbors to explicitly correct perplexity estimates, yielding a robust \emph{Sample Cleanliness Score}. This metric uncovers absolute sample cleanliness regardless of the corpus’s noise level. Extensive experiments demonstrate our method’s superior noise-detection capabilities and show that its final ICL performance is comparable to that of a fully clean demonstration corpus. Moreover, our approach remains robust even when noise ratios are extremely high.

Dual Debiasing for Noisy In-Context Learning for Text Generation

Recent advancements in deep learning have significantly enhanced the efficiency and accuracy of natural language processing (NLP) tasks. However, these models often require substantial computational resources, which remains a major drawback. Reducing the complexity of deep learning architectures, and exploring simpler yet effective approaches can lead to cost-efficient NLP solutions. This is also a step towards explainable AI, i.e., uncovering how a particular task is carried out. For this analysis, we chose the task of hate speech detection. We address hate speech detection by introducing a model that employs a weighted sum of valence, arousal, and dominance (VAD) scores for classification. To determine the optimal weights and classification strategies, we analyze hate speech and non-hate speech words based on both their individual and summed VAD-values. Our experimental results demonstrate that this straightforward approach can compete with state-of-the-art neural network methods, including GPT-based models, in detecting hate speech.

A Step Towards Explainable Hate Speech Detection

Knowledge Graphs represent real-world entities and the relationships between them. Multilingual Knowledge Graph Construction (mKGC) refers to the task of automatically constructing or predicting missing entities and links for knowledge graphs in a multilingual setting. In this work, we reformulate the mKGC task as a Question Answering (QA) task and introduce mRAKL: a Retrieval-Augmented Generation (RAG) based system to perform mKGC. We achieve this by using the head entity and linking relation in a question, and having our model predict the tail entity as an answer. Our experiments focus primarily on two low-resourced languages: Tigrinya and Amharic. We experiment with using higher-resourced languages, Arabic and English, to utilize cross-lingual transfer for mKGC. With a BM25 retriever, we find that the RAG-based approach improves performance over a no-context setting. Further, our ablation studies show that with an idealized retrieval system, mRAKL improves accuracy by up to 4.92 and 8.79 percentage points for Tigrinya and Amharic, respectively.

mRAKL: Multilingual Retrieval-Augmented Knowledge Graph Construction for Low-Resourced Languages

Recent success of large language models (LLMs) in diverse domains showcases their potential to revolutionize scientific fields, including drug editing. Traditional drug editing relies on iterative conversations with domain experts, refining the drug until the desired property is achieved. This interactive and iterative process mirrors the strengths of LLMs, making them well-suited for drug editing. *In existing works, LLMs edit each molecule independently without leveraging knowledge from past edits.* However, human experts develop intuition about effective modifications over time through historical experience; accumulating past knowledge is pivotal for human experts, and so it is for LLMs. *In this work, we propose RL-Guider---a reinforcement-learning-agent to provide suggestions to LLMs; it uses the rich information provided from evaluating editing results made by the LLM based on the recommendations to improve itself over time.* RL-Guider is the first work that leverages both the comprehensive "world-level" knowledge of LLMs and the knowledge accumulated from historical feedback. As a result, RL-Guider mitigates several shortcomings of existing approaches and demonstrates superior performance.

Downloads

Next from ACL 2025

Chain of Attack: Hide Your Intention through Multi-Turn Interrogation

Similar lecture

R-PRM: Reasoning-Driven Process Reward Modeling