Morocco

LLMs, while outperforming humans in a wide range of tasks, can still fail in unanticipated ways. We focus on two pervasive failure modes: (i) hallucinations, where models produce incorrect information about the world, and (ii) the low-resource effect, where the models show impressive performance in high-resource languages like English but the performance degrades significantly in low-resource languages like Bengali. We study the intersection of these issues and ask: do hallucination detectors suffer from the low-resource effect? We conduct experiments on five tasks across three domains (factual recall, STEM, and Humanities). Experiments with four LLMs and three hallucination detectors reveal a curious finding: As expected, the task accuracies in low-resource languages experience large drops (compared to English). However, the drop in hallucination detection accuracy is often several times smaller than the drop in task accuracy. Our finding suggests that even in low-resource languages, the internal mechanisms of LLMs might encode valuable signals about their uncertainty.

EACL 2026

Do LLM hallucination detectors suffer from low-resource effect?

model-internal artifacts

multilingual and cross-lingual evaluation

llms hallucination detection

low-resource languages

## Welcome to 2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics!
This year, the conference is in Mexico City. NAACL was actually already planned for Mexico City in 2021, but due to the pandemic the entire conference was moved online. This year, finally, we get to go! So it is my sincere pleasure to welcome you to Mexico City, whether in person or virtually. Having the conference in Mexico City is a good opportunity to emphasize that NAACL is our flagship conference for ACL members not only in North America but also in Central and South America, even though NAACL has been bearing “North” in its name. At this year’s conference, we have a theme to match, with a theme track on the Languages of Latin America to showcase the linguistic diversity of the region.
 
The opportunity to present at NAACL should not depend on a researcher’s travel budget, or their family status. This is why it is so important to make virtual participation at NAACL as good an experience as possible – but we want to also provide a good experience for in-person participants. As a community, we are still working out the best way to do that. This year at NAACL, we are trying out a big virtual poster session ahead of the conference, with the hope that this will make make for a lively and interactive experience. At the same time, we are reducing virtual oral presentations, which seem to be particularly tricky to make to work well. A big thanks to the NAACL program chairs and to Luciana Benotti for all their ideas and work to improve the virtual experience. And participants, virtual as well as in-person: Please let us know what worked for you and what didn’t, so we can continue to improve hybrid conferences. 
 
I have been lucky to work with many amazing people. Without their insight, dedication and patience, and without the many hours of work they put in, NAACL would not have been possible. A huge thank you to the program chairs Helena Gomez, Kevin Duh, and Steve Bethard – you are the best! 
 
Finally, I would like to thank all authors, invited speakers and panelists, area chairs and reviewers, the volunteers organizing and chairing sessions, and all attendees, in-person and virtual. Thank you for helping us make NAACL 2024 come to life. 
 
Welcome and hope you all enjoy the conference! 
Katrin Erk 
The University of Texas at Austin 
NAACL-2024 General Chair 
*You can read the full Welcome message in the [Conference Handbook (downloadable)](https://drive.google.com/file/d/1H1NvW0VASQjkSCYgw3mr-3l4yYDSxz6B/view?usp=sharing)*

To access the event page you need to register [**here.**](http://acl.swoogo.com/naacl2024) 
Your access to the event page is limited based on your registration type. If you registered for workshops only, you will gain access to full workshops content on the day of the workshops program.

Please register!

NAACL 2024

2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics

technical paper

#### *Message from the General Chair, Aline Villavicencio*
I’m delighted and honoured to welcome you to the 19th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2026), taking place in the beautiful city of Rabat, in Morocco, in March 24-29, 2026. EACL is the flagship European conference of the Association and EACL 2026 proudly continues our field’s tradition of excellence in scholarship, innovation, and inclusivity. I am deeply grateful to the many volunteers whose dedication, generosity, and tireless efforts have made this conference possible.
For the first time EACL is being hosted in the African continent. This is an important milestone for our community, and we are grateful to our Moroccan hosts for enabling this historic moment by bringing this edition of EACL to Rabat. We are also delighted that the Second Arabic NLP School is co-located with EACL. We hope attendees enjoy this wonderful opportunity to strengthen ties with the Computational Linguistics communities across the African continent. *[Read full message](https://drive.google.com/file/d/14NlmHvwM6fPJuMmOvVh7K0vtQbEyv3SZ/view?usp=sharing)* 

<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://underline.io/events/522/reception','_blank')">Go to Workshops and Tutorials Program</button></html>

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 EACL 2026". 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 are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.eacl.org/registration/) first.

**Online Registration Form**: https://acl.swoogo.com/eacl2026

Registration Required

Welcome to the 19th Conference of the European Chapter of the Association for Computational Linguistics (EACL). EACL 2026 will be held in Rabat, Morocco, from March 24–29, 2026. 

Large language models (LLMs) have demonstrated impressive few-shot in-context learning (ICL) abilities. Still, we show that they are sometimes prone to a `copying bias', where they copy answers from provided examples instead of learning the underlying patterns. In this work, we propose a novel and simple method to mitigate such copying bias. First, we create a synthetic task and use the Integrated Gradients method to identify neurons that prioritize copying over generalization. We demonstrate that pruning these neurons consistently improves performance across a diverse set of ICL tasks, including both single-token and multi-token scenarios, while maintaining or even improving the model's general capabilities. We also show that our method is applicable across various LLM architectures, including Transformers and State-Space Models, without requiring modifications. In our analysis, we adopt a task-recognition perspective on ICL and examine task vectors (Hendel et al., 2023) induced by the model. We find that pruning enhances the quality of these vectors, suggesting that the pruned neurons previously hindered effective task recognition.

Mitigating Copy Bias in In-Context Learning through Neuron Pruning

Large language models (LLMs) increasingly rely on external tools and APIs to execute complex tasks specified in natural language. Evaluating such tool-calling capabilities in realistic enterprise settings is challenging: APIs are often proprietary, heterogeneous, and difficult to share, limiting reproducible benchmarks. To address this, we introduce Live API Bench, a comprehensive benchmark constructed by transforming NL2SQL datasets into interactive API environments. Our pipeline converts SQL queries from BIRD-SQL into executable API sequences across three formulations—SLOT, SEL, and REST—covering minimal general-purpose operations, domain-specific multi-step tasks, and function-oriented RESTful interactions, respectively. The benchmark spans 11 databases with over 2,500 invocable tools, paired with human-authored queries, ground-truth API sequences, and verified final answers. Live API Bench enables systematic evaluation of core challenges in tool use, including error handling, sequential reasoning, parameter generation, response parsing, and robustness across diverse domains. We evaluate 10 LLMs and 4 ReACT agents, observing low task completion rates (7–47%), which improve modestly to 50% under interactive agent settings, highlighting substantial scope for improving LLM tool-calling performance. We release all code and data associated with this paper.

Live API-Bench: 2500+ Live APIs for Testing Multi-Step Tool Calling

This paper presents a novel task of extracting Latin fragments from mixed-language historical documents with varied layouts. We benchmark and evaluate the performance of large foundation models against a multimodal dataset of 724 annotated pages. The results demonstrate that reliable Latin detection with contemporary models is achievable. Our study provides the first comprehensive analysis of these models' capabilities and limits for this task.

Detecting Latin in Historical Books with Large Language Models: A Multimodal Benchmark

Multilingual domain adaptation (ML-DA) is widely used to learn new domain knowledge across languages into large language models (LLMs). Although many methods have been proposed to improve domain adaptation, the mechanisms of multilingual knowledge acquisition, how domain knowledge is learned within a language and transferred across languages, remain underexplored. This gap leads to suboptimal performance, particularly in low-resource settings. This work examines the learning dynamics of LLMs during ML-DA. Because prior ML-DA studies often train and evaluate on datasets with mismatched knowledge coverage, we propose \textbf{AdaXEval}, an adaptive evaluation method that builds multiple-choice QA datasets from the same bilingual domain corpus used for training, thereby directly studying multilingual knowledge acquisition. Through continual training of LLMs with diverse data recipes, we track how LLMs acquire domain facts and pinpoint the mechanism behind the transformation process from domain training data to knowledge. Our experiments on a 13B English-Japanese bilingual LLM reveal that cross-lingual transfer remains challenging despite a high-quality bilingual corpus. The code has been released.

Tracing Multilingual Knowledge Acquisition Dynamics in Domain Adaptation: A Case Study of English-Japanese Biomedical Adaptation

Speculative generation has emerged as a promising technique to accelerate inference in large language models (LLMs) by leveraging parallelism to verify multiple draft tokens simultaneously. However, the fundamental limits on the achievable speedup remain poorly understood. In this work, we establish the first ``tight'' lower bounds on the runtime of any deterministic speculative generation algorithm. This is achieved by drawing a parallel between the token generation process and branching random walks, which allows us to analyze the optimal draft tree selection problem. We prove, under basic assumptions, that the expected number of tokens successfully predicted per speculative iteration is bounded as E[X] (+ (2)) (B)/ 2 + O(1), where B is the verifier's batch size, is the expected entropy of the verifier's output distribution, and (2) is this entropy's second moment. This result provides new insights into the limits of parallel token generation, and could guide the design of future speculative decoding systems. Empirical evaluations on Llama models validate our theoretical predictions, confirming the tightness of our bounds in practical settings.

Speculative Decoding Speed-of-Light: Optimal Lower Bounds via Branching Random Walks

Existing datasets for multimodal table understanding, such as MMTab, primarily provide short factual answers without explicit multi-step reasoning supervision. Models trained on these datasets often generate brief responses that offers insufficient accuracy and limited interpretability into how these models arrive at the final answer. We introduce CoReTab, a code-driven reasoning framework that produces scalable, interpretable, and automatically verifiable annotations by coupling natural-language explanations with executable Python code. Using the CoReTab framework, we curate a dataset of 115K verified samples averaging 529 tokens per response and fine-tune open-source MLLMs through a three-stage pipeline. We evaluate the resulting model trained on CoReTab across 17 MMTab benchmarks spanning table question answering, fact verification, and table structure understanding. Our model achieves significant gains of +6.5%, +6%, and +25.6%, respectively, over MMTab-trained baselines, while producing transparent and verifiable reasoning traces. These results establish CoReTab as a robust and generalizable supervision framework for improving multi-step reasoning in multimodal table understanding.

CoReTab: Improving Multimodal Table Understanding with Code-driven Reasoning

Integrating external knowledge into Large Language Models (LLMs) is crucial for many real-world applications, yet current methods like Retrieval-Augmented Generation (RAG) face limitations with broad, multi-source queries, while long-context models are computationally prohibitive. We introduce CacheNotes: Task-Aware Key-Value Cache Compression. Given a task description and a corpus, CacheNotes first generates a sequence of Compression-Planning-Tokens (CPTs), an offline task-focused distillation pass that identifies and organizes key information from the corpus. These CPTs are then used to guide a one-time compression of the corpus into a compact, reusable KV cache, which is then used alone at inference time to efficiently answer diverse, reasoning-intensive queries, eliminating repeated retrieval or context expansion. Experiments on LongBench show that CacheNotes outperforms RAG by over 8 F1 points on Question-Answering tasks at a 0.95 compression and reduces latency by over 4x. On RULER, it surpasses previous query-agnostic compression methods by 55 points, narrowing the gap to query-aware compression approaches. Additional results on real-world enterprise and synthetic datasets demonstrate its strong performance on multi-hop and broad-coverage queries.

CacheNotes: Task-Aware Key-Value Cache Compression for Reasoning-Intensive Knowledge Tasks

Hyperscaling of data and parameter count in LLMs is yielding diminishing improvement, especially when weighed against training costs, underlining a growing need for more efficient finetuning and inference without sacrificing performance. This is especially so for multimodal models (MLMs), where the overhead of processing multimodal tokens alongside language data often limits the practical viability of these systems. Parallely, recent work has uncovered implicit cross-modal alignment in the deeper layers of large MLMs, improving our understanding of how MLMs process and encode information. Motivated by this, and our observation that MLMs naturally defer most cross-modal token interactions to deeper layers of the model, we propose a simple modification. Instead of concatenation with the language prompt at the start, we insert multimodal tokens directly into the middle, allowing them to entirely bypass the early layers. Our results with diverse modalities, (i) LLaVA & BLIP for vision, (ii) LTU for audio, and (iii) MoLCA for molecular data, and broad model sizes, starting from 350M and going upto 13B parameters, indicate that our method reduces both training and inference costs, while at least preserving, if not surpassing the performance of existing baselines.

DeepInsert: Early Layer Bypass for Efficient and Performant Multimodal Understanding

Language models (LMs) have shown impressive reasoning capabilities across various domains. A fundamental question is the extent of their reasoning power. While recent studies show that LMs can solve NP-complete problems, their ability to handle PSPACE-complete problems remains underexplored. We investigate regex minimization as a PSPACE-complete challenge for LMs to address this issue. Regexes, formal expressions for regular languages, are widely used in NLP, SE or PL. There are several efficient tools to manipulate them with theoretical background. Inspired by this, we introduce the first benchmark for regex minimization containing over a million regexes paired with their minimal equivalents. Through extensive experiments with two LMs trained on our dataset and five open-source large language models (LLMs), we analyze how well LMs perform on PSPACE-complete problems, highlighting their capabilities of generalization and limitations in reasoning. To the best of our knowledge, this is the first study to systematically evaluate LM reasoning in regex minimization and establish a foundation for solving PSPACE-complete problems with LMs.

A Regex Minimization Benchmark: A PSPACE-Complete Challenge for Language Models

Communication across languages has been historically a "holy grail" in the context of AI. In a world with more than 7000 languages, AI still falls short in terms of coverage or equitably serving different languages. For decades we have focused on high-resource languages, since models have largely been resource-dependent. Today we are witnessing the evolution of techniques that can contribute to better represent a larger number of languages. We are kind of breaking the “digital desert”. In this talk, we will cover an overview of the path towards Omnilinguality in AI through the lens of the Machine Translation, starting with basic but complex definitions such as language or long-tail. MT is one of the most popular and well-explored applications in multilinguality. MT has been shown to scale to a massive number of languages with flagship specialised models covering from hundreds up to thousands of languages. Progress in MT has been largely driven by the scientific community effort of WMT that has put together competitions, benchmarks and constantly posing new challenges. Other key contributions have been the open-sourced evaluation datasets such as FLORES that cover hundreds of languages in the Wikipedia domain, and more recently BOUQuET that builds on top of non-English data, originally created and expanding to more domains. These initiatives are complemented with open-initiatives that allow the community to contribute with more languages. In my personal opinion, while MT has served as a platform to reach such progress, it may not always be the main one. Even if we have not yet solved this task, the rise of LLMs shows the opportunity to solve many tasks at once. MT will still be a relevant part of the puzzle by being a source of synthetic data generation and a representative evaluation in breadth. However, I will argue that Omniliguality should not be faced task-specifically anymore and it should be taken as a great arena to test more broadly advanced LLMs techniques and an opportunity to come up with new methods to massively scale general purpose evaluation.

Downloads

Next from EACL 2026

Mitigating Copy Bias in In-Context Learning through Neuron Pruning

Similar lecture

LLMs Are Few-Shot In-Context Low-Resource Language Learners

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from EACL 2026

Mitigating Copy Bias in In-Context Learning through Neuron Pruning

Similar lecture

LLMs Are Few-Shot In-Context Low-Resource Language Learners

Downloads