Austria

Evaluating Large Language Models (LLMs) requires effective methods to assess semantic consistency across multiple reversible transformations. Traditional self-consistency methods often fail to capture subtle semantic errors in multi-step tasks. We introduce ConsistencyChecker, a tree-based evaluation framework that measures LLMs&#39; ability to preserve semantic consistency during reversible transformation processes, sidestepping benchmark data contamination issues. Our approach constructs self-consistency trees where nodes represent text states after transformations (e.g., translation, code modification, paraphrasing) and edges represent pairs of opposite transformations. By analyzing semantic preservation between nodes at different tree depths, ConsistencyChecker quantifies model reliability without requiring manually annotated reference data. Experiments demonstrate that ConsistencyChecker reliably measures generalization abilities across models from 1.5B to 72B parameters. On translation tasks, GPT-4o Mini achieves the highest L3 consistency score of 98.0%. For code generation, Qwen 2.5 32B leads with 85.1% semantic consistency at L3. Results show Pearson correlation greater than 0.7 between our embedding-based scores and WMT 2024 rankings on 4 out of 5 shared language pairs, validating the method&#39;s effectiveness for benchmarking LLM performance without constructing new datasets.

ACL 2025

ConsistencyChecker: Tree-based Evaluation of LLM Generalization Capabilities

nterpretability and analysis of models for nlp

machine learning for nlp

resources and evaluation

### 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!

ACL 2024

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.

Evaluating Large Language Models (LLMs) requires effective methods to assess semantic consistency across multiple reversible transformations. Traditional self-consistency methods often fail to capture subtle semantic errors in multi-step tasks. We introduce ConsistencyChecker, a tree-based evaluation framework that measures LLMs' ability to preserve semantic consistency during reversible transformation processes, sidestepping benchmark data contamination issues. Our approach constructs self-consistency trees where nodes represent text states after transformations (e.g., translation, code modification, paraphrasing) and edges represent pairs of opposite transformations. By analyzing semantic preservation between nodes at different tree depths, ConsistencyChecker quantifies model reliability without requiring manually annotated reference data. Experiments demonstrate that ConsistencyChecker reliably measures generalization abilities across models from 1.5B to 72B parameters. On translation tasks, GPT-4o Mini achieves the highest L3 consistency score of 98.0%. For code generation, Qwen 2.5 32B leads with 85.1% semantic consistency at L3. Results show Pearson correlation greater than 0.7 between our embedding-based scores and WMT 2024 rankings on 4 out of 5 shared language pairs, validating the method's effectiveness for benchmarking LLM performance without constructing new datasets.

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)

Extracting spatial information from text is crucial for applications in geographic information systems (GIS), smart cities, disaster prevention, and logistics planning, helping decision-makers better understand geographic distributions and trends. However, the complex and unstructured nature of geographic expressions in text—including variations in place names, vague directions, and implicit spatial relationships—poses significant challenges for traditional methods. To overcome these challenges, we introduce SpatialWebAgent, an automated agent system that employs large language models (LLMs) to extract, standardize, and ground spatial information from natural language onto maps. SpatialWebAgent addresses the issues posed by diverse and ambiguous geographic expressions—ranging from varying place names and vague directional cues to implicit spatial relationships that require flexible combinations of localization functions—by leveraging the deep semantic understanding of LLMs. Our system extracts different types of spatial entities, then applies a series of geospatial functions to convert them into precise coordinates via geocoding APIs, and finally visualizes the results on interactive maps. Our demo is available at: https://sites.google.com/view/SpatialWebAgent.

SpatialWebAgent: Leveraging Large Language Models for Automated Spatial Information Extraction and Map Grounding

Large language models (LLMs) commonly struggle with specialized or emerging topics which are rarely seen in the training corpus. Graph-based retrieval-augmented generation (GraphRAG) addresses this by structuring domain knowledge as a graph for dynamic retrieval. However, existing pipelines involve complex engineering workflows, making it difficult to isolate the impact of individual components. It is also challenging to evaluate the retrieval effectiveness due to the overlap between the pretraining and evaluation datasets. In this work, we introduce ROGRAG, a Robustly Optimized GraphRAG framework. Specifically, we propose a multi-stage retrieval mechanism that integrates dual-level with logic form methods to improve retrieval robustness without increasing computational cost. To further refine the system, we incorporate various result verification methods and adopt an incremental database construction approach. Through extensive ablation experiments, we rigorously assess the effectiveness of each component. Our implementation includes comparative experiments on SeedBench, where Qwen2.5-7B-Instruct initially underperformed. ROGRAG significantly improves the score from 60.0% to 75.0% and outperforms mainstream methods. Experiments on domain-specific datasets reveal that dual-level retrieval enhances fuzzy matching, while logic form retrieval improves structured reasoning, highlighting the importance of adaptive retrieval. ROGRAG is deployed on an online research platform https://seedllm.org.cn and released as an open-source resource https://github.com/tpoisonooo/ROGRAG. It supports easy installation with pip and a demo video https://www.youtube.com/watch?v=NL1_cuIiX-w is provided.

ROGRAG: A Robustly Optimized GraphRAG Framework

The rapid expansion of memes on social media has highlighted the urgent need for effective approaches to detect harmful content. However, traditional data-driven approaches struggle to detect new memes due to their evolving nature and the lack of up-to-date annotated data. To address this issue, we propose MIND, a multi-agent framework for zero-shot harmful meme detection that does not rely on annotated data. MIND implements three key strategies: 1) We retrieve similar memes from an unannotated reference set to provide contextual information. 2) We propose a bi-directional insight derivation mechanism to extract a comprehensive understanding of similar memes. 3) We then employ a multi-agent debate mechanism to ensure robust decision-making through reasoned arbitration. Extensive experiments on three meme datasets demonstrate that our proposed framework not only outperforms existing zero-shot approaches but also shows strong generalization across different model architectures and parameter scales, providing a scalable solution for harmful meme detection.

MIND: A Multi-agent Framework for Zero-shot Harmful Meme Detection

As language models regularly make mistakes when solving math problems, automated identification of errors in the reasoning process becomes increasingly significant for their scalable oversight. In this paper, we introduce ProcessBench for measuring the ability to identify erroneous steps in mathematical reasoning. It consists of 3,400 test cases, primarily focused on competition- and Olympiad-level math problems. Each test case contains a step-by-step solution with error location annotated by human experts. Models are required to identify the earliest step that contains an error, or conclude that all steps are correct. We conduct extensive evaluation on ProcessBench, involving two types of models: process reward models (PRMs) and critic models, where for the latter we prompt general language models to critique each solution step by step. We draw two main observations: (1) Existing PRMs typically fail to generalize to more challenging math problems beyond GSM8K and MATH. They underperform both critic models (i.e., prompted general language models) and our own trained PRM that is straightforwardly fine-tuned on the PRM800K dataset. (2) The best open-source model, QwQ-32B-Preview, has demonstrated the critique capability competitive with the proprietary model GPT-4o, despite that it still lags behind the reasoning-specialized o1-mini. We hope ProcessBench can foster future research in reasoning process assessment, paving the way toward scalable oversight of language models.

ProcessBench: Identifying Process Errors in Mathematical Reasoning

Given the high computational cost of preference alignment training of large language models (LLMs), exploring efficient methods to reduce the training overhead remains an important and compelling research problem. Motivated by the observation that alignment training typically involves only small parameter changes without injecting new knowledge into models, we propose a straightforward method called ExPO (model extrapolation) to expedite LLMs' alignment with human preferences. Given a partially-trained model and its initial SFT checkpoint, ExPO improves the implicit optimization objective of alignment training by simply amplifying the parameter change based on a first-order approximation, without any additional training overhead. Through controlled experiments, we demonstrate that ExPO boosts a DPO model trained with only 20% steps to outperform the fully-trained one. Moreover, we show that ExPO notably improves existing open-source LLMs (ranging from 1.8B to 70B parameters) on the leading AlpacaEval 2.0 and MT-Bench benchmarks, which highlights ExPO's broader utility in efficiently enhancing LLM alignment.

Model Extrapolation Expedites Alignment

Stance detection is a pivotal task in Natural Language Processing (NLP), identifying textual attitudes toward various targets. Despite advances in using Large Language Models (LLMs), challenges persist due to hallucination-models generating plausible yet inaccurate content. Addressing these challenges, we introduce MPVStance, a framework that incorporates Multi-Perspective Verification (MPV) with Retrieval-Augmented Generation (RAG) across a structured five-step verification process. Our method enhances stance detection by rigorously validating each response from factual accuracy, logical consistency, contextual relevance, and other perspectives. Extensive testing on the SemEval-2016 and VAST datasets, including scenarios that challenge existing methods and comprehensive ablation studies, demonstrates that MPVStance significantly outperforms current models. It effectively mitigates hallucination issues and sets new benchmarks for reliability and accuracy in stance detection, particularly in zero-shot, few-shot, and challenging scenarios.

MPVStance: Mitigating Hallucinations in Stance Detection with Multi-Perspective Verification

Catastrophic forgetting remains a formidable obstacle to building an omniscient model in large language models (LLMs). Despite the pioneering research on task-level forgetting in LLM fine-tuning, there is scant focus on forgetting during pre-training. We systematically explored the existence and measurement of forgetting in pre-training, questioning traditional metrics such as perplexity (PPL) and introducing new metrics to better detect entity memory retention. Based on our revised assessment of forgetting metrics, we explored low-cost, straightforward methods to mitigate forgetting during the pre-training phase. In addition, we carefully analyzed the learning curves, offering insights into the dynamics of forgetting. Extensive evaluations and analyses on forgetting of pre-training could facilitate future research on LLMs.

Exploring Forgetting in Large Language Model Pre-Training

Machine-generated Text (MGT) detection is crucial for regulating and attributing online texts. While the existing MGT detectors achieve strong performance, they remain vulnerable to simple perturbations and adversarial attacks. To build an effective defense against malicious perturbations, we view MGT detection from a threat modeling perspective, that is, analyzing the model’s vulnerability from an adversary’s point of view and exploring effective mitigations. To this end, we introduce an adversarial framework for training a robust MGT detector, named GREedy Adversary PromoTed DefendER (GREATER). The GREATER consists of two key components: an adversary GREATER-A and a detector GREATER-D. The GREATER-D learns to defend against the adversarial attack from GREATER-A and generalizes the defense to other attacks. GREATER-A identifies and perturbs the critical tokens in embedding space, along with greedy search and pruning to generate stealthy and disruptive adversarial examples. Besides, we update the GREATER-A and GREATER-D synchronously, encouraging the GREATER-D to generalize its defense to different attacks and varying attack intensities. Our experimental results across 10 text perturbation strategies and 6 adversarial attacks show that our GREATER-D reduces the Attack Success Rate (ASR) by 0.67% compared with SOTA defense methods while our GREATER-A is demonstrated to be more effective and efficient than SOTA attack approaches. Codes and dataset are available in https://github.com/Liyuuuu111/GREATER.

Iron Sharpens Iron: Defending Against Attacks in Machine-Generated Text Detection with Adversarial Training

Training large language models (LLMs) poses challenges due to their massive scale and heterogeneous architectures. While adaptive optimizers like AdamW help address gradient variations, they still struggle with efficient and effective parameter-wise learning rate estimation, resulting in training instability, slow convergence, and poor compatibility with parameter-efficient fine-tuning (PEFT) techniques. This work introduces Scaling with Gradient Grouping (SGG), an optimizer wrapper that improves adaptive learning rate estimation by dynamic grouping and group-specific scaling. SGG first groups gradient statistics in each layer into clusters and then applies cluster-specific scaling to calibrate learning rates for each parameter, thus imposing collective group-wise constraints while maintaining precise per-parameter adaptation. Experiments on diverse (M)LLM benchmarks show that SGG integrates seamlessly with existing optimizers, and offers consistent gains and faster convergence over baselines, with various model sizes. Its stability across varying batch sizes and learning rates establishes SGG as a robust choice for LLM optimization.

Taming LLMs with Gradient Grouping

Language models hold incredible promise for enabling scientific discovery by synthesizing massive research corpora. Many complex scientific research questions have multiple plausible answers, each supported by evidence of varying strength. However, existing language models lack the capability to quantitatively and faithfully compare answer plausibility in terms of supporting evidence. To address this, we introduce Retrieve to Explain (R2E), a retrieval-based model that scores and ranks all possible answers to a research question based on evidence retrieved from a document corpus. The architecture represents each answer only in terms of its supporting evidence, with the answer itself masked. This allows us to extend feature attribution methods such as Shapley values, to transparently attribute answer scores to supporting evidence at inference time. The architecture also allows incorporation of new evidence without retraining, including non-textual data modalities templated into natural language. We developed R2E for the challenging scientific discovery task of drug target identification, a human-in-the-loop process where failures are extremely costly and explainability paramount. When predicting whether drug targets will subsequently be confirmed as efficacious in clinical trials, R2E not only matches non-explainable literature-based models but also surpasses a genetics-based target identification approach used throughout the pharmaceutical industry.

Downloads

Next from ACL 2025

SpatialWebAgent: Leveraging Large Language Models for Automated Spatial Information Extraction and Map Grounding

Similar lecture

Enhancing Semantic Consistency of Large Language Models through Model Editing: An Interpretability-Oriented Approach