Austria

Being able to thoroughly assess massive multi-task language understanding (MMLU) capabilities is essential for advancing the applicability of multilingual language models. However, preparing such benchmarks in high quality native language is often costly and therefore limits the representativeness of evaluation datasets. While recent efforts focused on building more inclusive MMLU benchmarks, these are conventionally built using machine translation from high-resource languages, which may introduce errors and fail to account for the linguistic and cultural intricacies of the target languages. In this paper, we address the lack of native language MMLU benchmark especially in the under-represented Turkic language family with distinct morphosyntactic and cultural characteristics. We propose two benchmarks for Turkic language MMLU: TUMLU is a comprehensive, multilingual, and natively developed language understanding benchmark specifically designed for Turkic languages. It consists of middle- and high-school level questions spanning 11 academic subjects in Azerbaijani, Crimean Tatar, Karakalpak, Kazakh, Kyrgyz, Tatar, Turkish, Uyghur, and Uzbek. We also present TUMLU-mini, a more concise, balanced, and manually verified subset of the dataset. Using this dataset, we systematically evaluate a diverse range of open and proprietary multilingual large language models (LLMs), including Claude, Gemini, GPT, and LLaMA, offering an in-depth analysis of their performance across different languages, subjects, and alphabets. To promote further research and development in multilingual language understanding, we release TUMLU-mini and all corresponding evaluation scripts.

ACL 2025

TUMLU: A Unified and Native Language Understanding Benchmark for Turkic Languages

multilingual qa

multilingual corpora

datasets for low resource languages

multilingual evaluation

### 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)

poster

The lack of data transparency in Large Language Models (LLMs) has highlighted the importance of Membership Inference Attack (MIA), which differentiates trained (member) and untrained (non-member) data. Though it shows success in previous studies, recent research reported a near-random performance in different settings, highlighting a significant performance inconsistency. We assume that a single setting doesn't represent the distribution of the vast corpora, causing members and non-members with different distributions to be sampled and causing inconsistency. In this study, instead of a single setting, we statistically revisit MIA methods from various settings with thousands of experiments for each MIA method, along with study in text feature, embedding, threshold decision, and decoding dynamics of members and non-members. We found that (1) MIA performance improves with model size and varies with domains, while most methods do not statistically outperform baselines, (2) Though MIA performance is generally low, a notable amount of differentiable member and non-member outliers exists and vary across MIA methods, (3) Deciding a threshold to separate members and non-members is an overlooked challenge, (4) Text dissimilarity and long text benefit MIA performance, (5) Differentiable or not is reflected in the LLM embedding, (6) Member and non-members show different decoding dynamics.

A Statistical and Multi-Perspective Revisiting of the Membership Inference Attack in Large Language Models

Supervised Fine-Tuning (SFT) with translated instruction data effectively adapts Large Language Models (LLMs) from English to non-English languages. We introduce Cross-Lingual Continued Instruction Tuning (X-CIT), which fully leverages translation-based parallel instruction data to enhance cross-lingual adaptability. X-CIT emulates the human process of second language acquisition and is guided by Chomsky's Principles and Parameters Theory. It first fine-tunes the LLM on English instruction data to establish foundational capabilities (i.e. Principles), then continues with target language translation and customized chat-instruction data to adjust "parameters" specific to the target language. This chat-instruction data captures alignment information in translated parallel data, guiding the model to initially think and respond in its native language before transitioning to the target language. To further mimic human learning progression, we incorporate Self-Paced Learning (SPL) during continued training, allowing the model to advance from simple to complex tasks. Implemented on Llama-2-7B across five languages, X-CIT was evaluated against three objective benchmarks and an LLM-as-a-judge benchmark, improving the strongest baseline by an average of 1.97% and 8.2% in these two benchmarks, respectively.

From English to Second Language Mastery: Enhancing LLMs with Cross-Lingual Continued Instruction Tuning

Modeling latent representations in a hyperspherical space has proven effective for capturing directional similarities in high-dimensional text data, benefiting topic modeling. Variational autoencoder-based neural topic models (VAE-NTMs) commonly adopt the von Mises-Fisher prior to encode hyperspherical structure. However, VAE-NTMs often suffer from posterior collapse, where the KL divergence term in the objective function highly diminishes, leading to ineffective latent representations. To mitigate this issue while modeling hyperspherical structure in the latent space, we propose the Spherical Sliced Wasserstein Autoencoder for Topic Modeling (S2WTM). S2WTM employs a prior distribution supported on the unit hypersphere and leverages the Spherical Sliced-Wasserstein distance to align the aggregated posterior distribution with the prior. Experimental results demonstrate that S2WTM outperforms state-of-the-art topic models, generating more coherent and diverse topics while improving performance on downstream tasks.

S2WTM: Spherical Sliced-Wasserstein Autoencoder for Topic Modeling

Cross-Document Coreference Resolution (CDCR) aims to identify and group together mentions of a specific event or entity that occur across multiple documents. In contrast to the within-document tasks, in which event and entity mentions are linked by rich and coherent contexts, cross-document mentions lack such critical contexts, which presents a significant challenge in establishing connections among them. To address this issue, we introduce a novel task Cross-Document Discourse Coherence Enhancement (CD-DCE) to enhance the discourse coherence between two cross-document event or entity mentions. Specifically, CD-DCE first selects coherent texts and then adds them between two cross-document mentions to form a new coherent document. Subsequently, the coherent text is employed to represent the event or entity mentions and to resolve any coreferent mentions. Experimental results on the three popular datasets demonstrate that our proposed method outperforms several state-of-the-art baselines.

Employing Discourse Coherence Enhancement to Improve Cross-Document Event and Entity Coreference Resolution

Post-training is essential for enabling large language models (LLMs) to follow human instructions. However, its effectiveness depends on high-quality instruction data, which is challenging to obtain in the real world due to privacy concerns, data scarcity, and high annotation costs. To fill this gap, inspired by the recent success of using LLMs to simulate human society, we propose MATRIX, a multi-agent simulator that automatically generates diverse text-based scenarios, capturing a wide range of real-world human needs in a realistic and scalable manner. Leveraging these outputs, we introduce a novel scenario-driven instruction generator MATRIX-Gen for controllable and highly realistic data synthesis. Extensive experiments demonstrate that our framework effectively generates both general and domain-specific data. On AlpacaEval 2 and Arena-Hard benchmarks, Llama-3-8B-Base, post-trained on datasets synthesized by MATRIX-Gen with just 20K instruction-response pairs, outperforms Meta's Llama-3-8B-Instruct model, which was trained on over 10M pairs.

Synthesizing Post-Training Data for LLMs through Multi-Agent Simulation

Recent advancements in large language models (LLMs) have significantly improved the performance of multi-hop question answering (MHQA) systems. Despite the success of MHQA systems, the evaluation of MHQA is not deeply investigated. Existing evaluations mainly focus on comparing the final answers of the reasoning method and given ground-truths. We argue that the reasoning process should also be evaluated because wrong reasoning process can also lead to the correct final answers. Motivated by this, we propose a "Planner-Executor-Reasoner" (PER) architecture, which forms the core of the Plan-anchored Data Preprocessing (PER-DP) and the Plan-guided Multi-Hop QA (PER-QA).
The former provides the ground-truth of intermediate reasoning steps and final answers, and the latter offers them of a reasoning method. Moreover, we design a fine-grained evaluation metric called Plan-aligned Stepwise Evaluation (PSE), which evaluates the intermediate reasoning steps from two aspects: planning and solving. Extensive experiments on ten types of questions demonstrate competitive reasoning performance, improved explainability of the MHQA system, and uncover issues such as "fortuitous reasoning continuance" and "latent reasoning suspension" in RAG-based MHQA systems. Besides, we also demonstrate the potential of our approach in data contamination scenarios.

Beyond the Answer: Advancing Multi-Hop QA with Fine-Grained Graph Reasoning and Evaluation

This paper investigates the flow of factual information in Mamba State-Space Model (SSM)-based language models. We rely on theoretical and empirical connections to Transformer-based architectures and their attention mechanisms. Exploiting this relationship, we adapt attentional interpretability techniques originally developed for Transformers—specifically, the Attention Knockout methodology—to both Mamba-1 and Mamba-2. Using them we trace how information is transmitted and localized across tokens and layers, revealing patterns of subject-token information emergence and layer-wise dynamics. Notably, some phenomena vary between mamba models and Transformer based models, while others appear universally across all models inspected—hinting that these may be inherent to LLMs in general. By further leveraging Mamba’s structured factorization, we disentangle how distinct “features” either enable token-to-token information exchange or enrich individual tokens, thus offering a unified lens to understand Mamba internal operations.

Mamba Knockout for Unraveling Factual Information Flow

Recent remarkable advances in Large Language Models (LLMs) have led to innovations in various domains such as education, healthcare, and finance, while also raising serious concerns that they can be easily misused for malicious purposes. Most previous research has focused primarily on observing how jailbreak attack techniques bypass safety mechanisms like Reinforcement Learning through Human Feedback (RLHF). However, whether there are neurons within LLMs that directly govern aggression has not been sufficiently investigated. To fill this gap, this study identifies specific neurons (“aggression neurons”) closely related to the expression of aggression and systematically analyzes how manipulating them affects the model’s overall aggression. Specifically, using a large-scale synthetic text corpus (aggressive and non-aggressive), we measure the activation frequency of each neuron, then apply masking and activation techniques to quantitatively evaluate changes in aggression by layer and by manipulation ratio. Experimental results show that, in all models, manipulating only a small number of neurons can increase aggression by up to 33%, and the effect is even more extreme when aggression neurons are concentrated in certain layers. Moreover, even models of the same scale exhibit nonlinear changes in aggression patterns, suggesting that simple external safety measures alone may not be sufficient for complete defense.

Small Changes, Big Impact: How Manipulating a Few Neurons Can Drastically Alter LLM Aggression

This paper explores a novel perspective to speech quality assessment by leveraging natural language descriptions, offering richer, more nuanced insights than traditional numerical scoring methods. Natural language feedback provides instructive recommendations and detailed evaluations, yet existing datasets lack the comprehensive annotations needed for this approach. To bridge this gap, we introduce QualiSpeech, a comprehensive low-level speech quality assessment dataset encompassing 11 key aspects and detailed natural language comments that include reasoning and contextual insights. Additionally, we propose the QualiSpeech Benchmark to evaluate the low-level speech understanding capabilities of auditory large language models (LLMs). Experimental results demonstrate that finetuned auditory LLMs can reliably generate detailed descriptions of noise and distortion, effectively identifying their types and temporal characteristics. The results further highlight the potential for incorporating reasoning to enhance the accuracy and reliability of quality assessments. The dataset can be found at https://huggingface.co/datasets/tsinghua-ee/QualiSpeech.

QualiSpeech: A Speech Quality Assessment Dataset with Natural Language Reasoning and Descriptions

Instruction tuning for large language models (LLMs) can drive them to produce results consistent with human goals in specific downstream tasks. However, the process of continual instruction tuning (CIT) for LLMs may bring about the catastrophic forgetting (CF) problem, where previously learned abilities are degraded. Recent methods try to alleviate the CF problem by modifying models or replaying data, which may only remember the surface-level pattern of instructions and get confused on held-out tasks. In this paper, we propose a novel continual instruction tuning method based on Key-part Information Gain (KPIG). Our method computes the information gain on masked parts to dynamically replay data and refine the training objective, which enables LLMs to capture task-aware information relevant to the correct response and alleviate overfitting to general descriptions in instructions. In addition, we propose two metrics, P-score and V-score, to measure the generalization and instruction-following abilities of LLMs. Experiments demonstrate our method achieves superior performance on both seen and held-out tasks.

Downloads

Next from ACL 2025

A Statistical and Multi-Perspective Revisiting of the Membership Inference Attack in Large Language Models

Similar lecture

Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation