Thailand

As diverse linguistic communities and users adopt Large Language Models (LLMs), assessing their safety across languages becomes critical. Despite ongoing efforts to align these models with safe and ethical guidelines, they can still be induced into unsafe behavior with jailbreaking, a technique in which models are prompted to act outside their operational guidelines. What research has been conducted on these vulnerabilities was predominantly on English, limiting the understanding of LLM behavior in other languages. We address this gap by investigating Many-Shot Jailbreaking (MSJ) in Italian, underscoring the importance of understanding LLM behavior in different languages. We base our analysis on a newly created Italian dataset to identify unique safety vulnerabilities in 4 families of open-source LLMs. We find that the models exhibit unsafe behaviors even with minimal exposure to harmful prompts, and--more alarmingly--this tendency rapidly escalates with more demonstrations.

ACL 2024

Compromesso! Italian Many-Shot Jailbreaks undermine the safety of Large Language Models

model vulnerabilities

italian language processing

multilingual model safety

### 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**. 

ACL 2025

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

poster

### 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!

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.

Transformer model has been a de-facto standard in natural language processing. Its adaptations in other fields such as computer vision showed promising results that this architecture is a powerful neural network in representation learning regardless of the data type. This recent success has led to research in multimodal Large Language Model (LLM), which enabled us to new types of tasks and applications with multiple data types. However, multimodal LLM in the biomedical domain is primarily limited to images, text, and/or sequence data. Here I propose to work on multimodal LLM architecture for biomedical graphs such as protein structure and chemical molecules. The research hypothesis is based on the fact that clinicians and researchers in computational biology and clinical research take advantage of various information for their decision-making process. Therefore, an AI model being able to handle multiple data types should boost its ability to use diverse knowledge for improved performances in clinical applications.

Foundation Model for Biomedical Graphs: Integrating Knowledge Graphs and Protein Structures to Large Language Models

Question answering involves creating answers to questions. With the growth of large language models, the ability of question-answering systems has dramatically improved. However, there is a lack of Vietnamese abstractive question-answering datasets, especially in the medical domain. Therefore, this research aims to mitigate this gap by introducing ViMedAQA. This **Vi**etnamese **Med**ical **A**bstractive **Q**uestion-**A**nswering dataset covers four topics in the Vietnamese medical domain, including body parts, disease, drugs and medicine. Additionally, the empirical results on the proposed dataset examine the capability of the large language models in the Vietnamese medical domain, including reasoning, memorizing and awareness of essential information.

ViMedAQA: A Vietnamese Medical Abstractive Question-Answering Dataset and Findings of Large Language Model

Geoparsing, the task of assigning coordinates to locations extracted from free text, is invaluable in enabling us to place locations in time and space. In the historical domain, many geoparsing corpora are from large news collections. We examine the Svoboda Diaries, a small historical corpus written primarily in English, with many location names in transliterated Arabic. We develop a pipeline employing named entity recognition for geotagging, and a map-based generate-and-rank approach incorporating name augmentation and clustering of location context words for geocoding. Our system outperforms existing map-based geoparsers in terms of correct location identification and lowest mean distance error. As location names may vary from those in knowledge bases, we find that augmented candidate generation is instrumental in the system's performance. Among our candidate generation methods, the generation of translated names contributed the most to increased location matches in the knowledge base. Our main contribution is proposing an integrated pipeline for geoparsing of historical corpora using augmented candidate location name generation and clustering methods -- an approach that can be generalized to other texts with foreign or non-standard spellings.

Basreh or Basra? Geoparsing Historical Locations in the Svoboda Diaries

Knowledge distillation allows smaller neural networks to emulate the performance of larger, teacher models with reduced computational demands. Traditional methods for Large Language Models (LLMs) often necessitate extensive fine-tuning, which limits their accessibility. To address this, we introduce Trace-of-Thought Prompting, a novel framework designed to distill critical reasoning capabilities from high-resource teacher models (over 8 billion parameters) to low-resource student models (up to 8 billion parameters). This approach leverages problem decomposition to enhance interpretability and facilitate human-in-the-loop interventions. Empirical evaluations on the GSM8K and MATH datasets show that student models achieve accuracy gains of up to 113\% on GSM8K and 20\% on MATH, with significant improvements particularly notable in smaller models like Llama 2 and Zephyr. Our results suggest a promising pathway for open-source, low-resource models to eventually serve as both students and teachers, potentially reducing our reliance on high-resource, proprietary models. Our code, featuring data analytics and testing scripts, is provided here: https://github.com/traceofthought/trace-of-thought-prompting/tree/main.

Trace-of-Thought Prompting: Investigating Prompt-Based Knowledge Distillation Through Question Decomposition

Neural networks (NN) perform well in diverse tasks, but sometimes produce nonsensical results to humans. Most NN models "solely" learn from (input, output) pairs, occasionally conflicting with human knowledge. Many studies indicate injecting human knowledge by reducing output constraints during training can improve model performance and reduce constraint violations. While there have been several attempts to compare different existing algorithms under the same programming framework, nonetheless, there has been no previous work that categorizes learning algorithms with output constraints in a unified manner. Our contributions are as follows: (1) We categorize the previous studies based on three axes: type of constraint loss used (e.g. probabilistic soft logic, REINFORCE), exploration strategy of constraint-violating examples, and integration mechanism of learning signals from main task and constraint. (2) We propose new algorithms to integrate the information of main task and constraint injection, inspired by continual-learning algorithms. (3) Furthermore, we propose the $H\beta$-score as a metric for considering the main task metric and constraint violation simultaneously. To provide a thorough analysis, we examine all the algorithms on three NLP tasks: natural language inference (NLI), synthetic transduction examples (STE), and semantic role labeling (SRL). We explore and reveal the key factors of various algorithms associated with achieving high $H\beta$-scores.

An Analysis under a Unified Formulation of Learning Algorithms with Output Constraints

The biomedical field relies on cost and time intensive systematic reviews of publications to enable practitioners to keep up to date with research. Impressive recent advances in large language models (LLMs) have made the task of automating at least part of the systematic review process feasible, but progress is slow. This paper identifies some factors that may have been holding research back, and proposes a new, enhanced dataset and prompting-based method for automatic synthesis generation, the most challenging step for automation. We test different models and types of information from and about biomedical studies for their usefulness in obtaining high-quality results. We find that, surprisingly, inclusion of paper abstracts generally worsens results. Instead, study summary information, and system instructions informed by domain knowledge are key to producing high-quality syntheses.

Beyond Abstracts: A New Dataset, Prompt Design Strategy and Method for Biomedical Synthesis Generation

Decoder-based large language models (LLMs) have shown high performance on many tasks in natural language processing. This is also true for sentence embedding learning, where a decoder-based model, PromptEOL, has achieved the best performance on semantic textual similarity (STS) tasks. However, PromptEOL requires a manually annotated natural language inference (NLI) dataset for fine-tuning. We aim to improve sentence embeddings without using large manually annotated datasets by automatically generating an NLI dataset with an LLM and using it for fine-tuning of PromptEOL. To achieve this, we explore methods of data generation suitable for sentence embedding learning in this study. Specifically, we will focus on automatic dataset generation through few-shot learning and explore the appropriate methods to leverage few-shot examples. Experimental results on the STS tasks demonstrate that our approach outperforms existing models in settings without large manually annotated datasets.

Improving Sentence Embeddings with Automatic Generation of Training Data Using Few-shot Examples

Research on understanding and generating diagrams has used vision models such as CLIP. However, it remains unclear whether these models accurately identify diagram attributes, such as node colors and shapes, along with edge colors and connection patterns. This study evaluates how well vision models recognize the diagram attributes by probing the model and retrieving diagrams using text queries. Experimental results showed that while vision models can recognize differences in node colors, shapes, and edge colors, they struggle to identify differences in edge connection patterns that play a pivotal role in the semantics of diagrams. Moreover, we revealed inadequate alignment between diagram attributes and language representations in the embedding space.

How Well Do Vision Models Encode Diagram Attributes?

Although Large Language Models (LLMs) have been trained using just the next token prediction objective, these have shown impressive performance on various tasks. Consequently, it has attracted research interests in this regard. While one line of work in the past has suggested that LLMs learn surface-level statistics from the dataset, another line of work emphasizes that the learned representations are effective for simulating the underlying world model, considering the causal relationship for the next token prediction. This phenomenon is often referred to as the emergence of a world model in sequence prediction tasks. Recent work has demonstrated this phenomenon in a simulated setting of board games like Othello and Chess. In this paper, we analyze the game of Checkers to find out the emergence of a world model in a language model. By training a GPT-style autoregressive language model using only the next character prediction objective, we find that the model does learn a world model representation of the board positions. We perform our analysis on two datasets: 1) from human gameplay and 2) synthetic, which comes from the checkers game tree. With multiple models trained with different layer sizes, we find that increasing the parameter size does help learn better world model representation decoded by linear probes.

CheckersGPT: Learning World Models through Language Modeling

Pre-training large language models faces significant memory challenges due to the large size of model weights. We propose STaged parameter-Efficient Pre-training (STEP), which combines ideas from parameter-efficient tuning and staged training. We conduct experiments on pre-training models of various sizes and demonstrate that STEP can achieve up to a 40.4\% reduction in maximum memory requirement compared to vanilla pre-training while maintaining comparable performance.

Downloads

Next from ACL 2024

Foundation Model for Biomedical Graphs: Integrating Knowledge Graphs and Protein Structures to Large Language Models

Similar lecture

What Really Matters in Many-Shot Attacks? An Empirical Study of Long-Context Vulnerabilities in LLMs

.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 ACL 2024

Foundation Model for Biomedical Graphs: Integrating Knowledge Graphs and Protein Structures to Large Language Models

Similar lecture

What Really Matters in Many-Shot Attacks? An Empirical Study of Long-Context Vulnerabilities in LLMs

Downloads