India

Large language models (LLMs) have shown significant progress in reasoning tasks. However, recent studies show that transformers and LLMs fail catastrophically once reasoning problems exceed modest complexity. We revisit these findings through the lens of large reasoning models (LRMs)
LLMs fine-tuned with incentives for step‑by‑step argumentation and self‑verification. LRM performance on graph and reasoning benchmarks such as **NLGraph** seem extraordinary, with some even claiming they are capable of generalized reasoning and innovation in reasoning-intensive fields such as mathematics, physics, medicine, and law. However, by more carefully scaling the complexity of reasoning problems, we show existing benchmarks actually have limited complexity. We develop a new dataset, the **Deep Reasoning Dataset (DeepRD)**, along with a generative process for producing unlimited examples of scalable complexity. We use this dataset to evaluate model performance on graph connectivity and natural language proof planning. We find that the performance of LRMs drop abruptly at sufficient complexity and do not generalize. We also relate our LRM results to the distributions of the complexities of large, real-world knowledge graphs, interaction graphs, and proof datasets. We find the majority of real-world examples fall inside the LRMs&#39; success regime, yet the long tails expose substantial failure potential. Our analysis highlights the near-term utility of LRMs while underscoring the need for new methods that generalize beyond the complexity of examples in the training distribution.

IJCNLP-AACL 2025

Reasoning Models Reason Well, Until They Don’t

proof planning

logical deduction

nlp datasets

benchmarking

reasoning

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**. 

ACL 2025

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

Large language models (LLMs) have shown significant progress in reasoning tasks. However, recent studies show that transformers and LLMs fail catastrophically once reasoning problems exceed modest complexity. We revisit these findings through the lens of large reasoning models (LRMs)
LLMs fine-tuned with incentives for step‑by‑step argumentation and self‑verification. LRM performance on graph and reasoning benchmarks such as **NLGraph** seem extraordinary, with some even claiming they are capable of generalized reasoning and innovation in reasoning-intensive fields such as mathematics, physics, medicine, and law. However, by more carefully scaling the complexity of reasoning problems, we show existing benchmarks actually have limited complexity. We develop a new dataset, the **Deep Reasoning Dataset (DeepRD)**, along with a generative process for producing unlimited examples of scalable complexity. We use this dataset to evaluate model performance on graph connectivity and natural language proof planning. We find that the performance of LRMs drop abruptly at sufficient complexity and do not generalize. We also relate our LRM results to the distributions of the complexities of large, real-world knowledge graphs, interaction graphs, and proof datasets. We find the majority of real-world examples fall inside the LRMs' success regime, yet the long tails expose substantial failure potential. Our analysis highlights the near-term utility of LRMs while underscoring the need for new methods that generalize beyond the complexity of examples in the training distribution.

technical paper

### Welcome to IJCNLP-AACL 2025! 
 It is a great honor to host this joint conference in Mumbai, India, from December 20 to 24, 2025. The joint conferences of IJCNLP and AACL are organized with alternating leadership in the Asia-Pacific region. The event is run by the Asian Federation of Natural Language Processing (AFNLP) in odd years, and by AACL in even years, while it is organized solely by ACL when the annual ACL meeting is held in the region. This year, the conference is primarily organized by AFNLP. 
*Kentaro Inui
MBZUAI, UAE
General Chair, IJCNLP-AACL 2025* 
Read full message and download the Conference Handbook [**here**](https://drive.google.com/file/d/1UTwxkAqSqI-GAoJC3wE1zZt5VP1Y8GX0/view?usp=sharing).

The 14th IJCNLP & 4th AACL will be held in Mumbai, India from December 20th to December 24th, 2025.

We investigate communication emerging in noisy environments with the goal of capturing the impact of message disruption on the emerged protocols. We implement two different noise mechanisms, inspired by the erasure and deletion channels studied in information theory, and simulate a referential game in a neural agent-based model with a variable message length channel. We leverage a stochastic evaluation setting to apply noise only after a message is sampled, which adds ecological validity and allows us to estimate information-theoretic measures of the emerged protocol directly from symbol probabilities. Contrary to our expectations, the emerged protocols do not become more redundant with the presence of noise; instead, we observe that certain levels of noise encourage the sender to produce more compositional messages, although the impact varies depending on the type of noise and input representation.

Noise May Drown Out Words but Foster Compositionality: The Advantage of the Erasure and Deletion Noisy Channels on Emergent Communication

The rapid proliferation of frontier model agents promises significant societal advances but also raises concerns about systemic risks arising from unsafe interactions.
Collusion to the disadvantage of others has been identified as a central form of undesirable agent cooperation.
The use of information hiding (steganography) in agent communications could render such collusion practically undetectable.
This underscores the need for investigations into the possibility of such behaviours emerging and the robustness corresponding countermeasures.
To investigate this problem we design two approaches -- a gradient-based reinforcement learning (GBRL) method and an in-context reinforcement learning (ICRL) method -- for reliably eliciting sophisticated LLM-generated linguistic text steganography.
We demonstrate, for the first time, that unintended steganographic collusion in LLMs can arise due to mispecified reward incentives during training.
Additionally, we find that standard mitigations --- both passive oversight of model outputs and active mitigation through communication paraphrasing --- are not fully effective at preventing this steganographic communication.
Our findings imply that (i) emergence of steganographic collusion is a plausible concern that should be monitored and researched, and (ii) preventing emergence may require innovation in mitigation techniques.

Hidden in Plain Text: Emergence & Mitigation of Steganographic Collusion in LLMs

State space models (SSMs) are a promising alternative to transformers for language modeling because they use fixed memory during inference. However, this fixed memory usage requires some information loss in the hidden state when processing long sequences. While prior work has studied the sequence length at which this information loss occurs, it does not characterize the types of information SSM language models (LMs) tend to forget. In this paper, we address this knowledge gap by identifying the types of tokens (e.g., parts of speech, named entities) and sequences (e.g., code, math problems) that are more frequently forgotten by SSM LMs. We achieve this by training an auto-encoder to reconstruct sequences from the SSM's hidden state, and measure information loss by comparing inputs with their reconstructions. We perform experiments using the Mamba family of SSM LMs (130M--1.4B) on sequences ranging from 4--256 tokens. Our results show significantly higher rates of information loss on math-related tokens (e.g., numbers, variables), mentions of organization entities, and alternative dialects to Standard American English. We then examine the frequency that these tokens appear in Mamba's pretraining data and find that less prevalent tokens tend to be the ones Mamba is most likely to forget. By identifying these patterns, our work provides clear direction for future research to develop methods that better control Mamba's ability to retain important information.

Characterizing Mamba’s Selective Memory using Auto-Encoders

Document Visual Question Answering (DocVQA) models often produce overconfident or ethically misaligned responses, especially under uncertainty. Existing models like LayoutLMv3, UDOP, and DONUT focus on accuracy but lack ethical calibration. We propose HonestVQA, a model-agnostic, self-supervised framework that aligns model confidence with correctness using weighted loss and contrastive learning. We introduce two new metrics—Honesty Score (H-Score) and Ethical Confidence Index (ECI)—to evaluate ethical alignment. HonestVQA improves accuracy and F1 by up to 4.3% across SpDocVQA, InfographicsVQA, and SROIE datasets, while reducing overconfidence. It also generalizes well across domains, achieving 78.9% accuracy and 76.1% F1-score.

The Confidence Paradox: Can LLM Know When It’s Wrong?

The output of large language models (LLMs) is unstable, due both to non-determinism of the decoding process as well as to prompt brittleness. While the intrinsic non-determinism of LLM generation may mimic existing uncertainty in human annotations through distributional shifts in outputs, it is largely assumed, yet unexplored, that the prompt brittleness effect is unique to LLMs. This raises the question: do human annotators show similar sensitivity to prompt changes? If so, should prompt brittleness in LLMs be considered problematic? One may alternatively hypothesize that prompt brittleness correctly reflects human annotation variances. To fill this research gap, we systematically compare the effects of prompt modifications on LLMs and identical instruction modifications for human annotators, focusing on the question of whether humans are similarly sensitive to prompt perturbations. To study this, we prompt both humans and LLMs for a set of text classification tasks conditioned on prompt variations. Our findings indicate that both humans and LLMs exhibit increased brittleness in response to specific types of prompt modifications, particularly those involving the substitution of alternative label sets or label formats. However, the distribution of human judgments is less affected by typographical errors and reversed label order than that of LLMs.

Are Humans as Brittle as Large Language Models?

The diversity across outputs generated by LLMs shapes perception of their quality and utility. High lexical diversity is often desirable, but there is no standard method to measure this property. Templated answer structures and ``canned'' responses across different documents are readily noticeable, but difficult to visualize across large corpora. This work aims to standardize measurement of text diversity. Specifically, we empirically investigate the convergent validity of existing scores across English texts, and release diversity, an open-source Python package (https://pypi.org/project/diversity/, https://github.com/cshaib/diversity) for measuring and extracting repetition in text. We also build a platform (https://ai-templates.app) based on diversity for users to interactively explore repetition in text. We find that fast compression algorithms capture information similar to what is measured by slow-to-compute $n$-gram overlap homogeneity scores. Further, a combination of measures---compression ratios, self-repetition of long $n$-grams, and Self-BLEU---are sufficient to report, as they have low mutual correlation with each other.

Standardizing the Measurement of Text Diversity: A Tool and Comparative Analysis

As cosmological simulations and their associated software become increasingly complex, physicists face the challenge of searching through vast amounts of literature and user manuals to extract simulation parameters from dense academic papers, each using different models, and formats. Translating these parameters into executable scripts remains a time-consuming and error-prone process. To improve efficiency in physics research and accelerate the cosmological simulation process, we introduce SimAgents, a multi-agent system designed to automate both parameter configuration from the literature and preliminary analysis for cosmology research. SimAgents is powered by specialized LLM agents capable of physics reasoning, simulation software validation, and tool execution. These agents collaborate through structured communication, ensuring that extracted parameters are physically meaningful, internally consistent, and software-compliant. We also construct a cosmological parameter extraction evaluation dataset by collecting over 40 simulations in published papers from Arxiv and leading journals that cover diverse simulation types. Experiments on the dataset demonstrate a strong performance of SimAgents, highlighting its effectiveness and potential to accelerate scientific research for physicists. Our demonstration video is available at: https://youtu.be/w1zLpm_CaWA. The complete system and dataset are publicly available at https://github.com/xwzhang98/SimAgents.

SimAgents: Bridging Literature and the Universe Via A Multi-Agent Large Language Model System

Despite the size of the field, stance detection has remained inaccessible to most researchers due to implementation barriers. Here we present a library that allows easy access to an end-to-end stance modelling solution. This library comes complete with everything needed to go from a corpus of documents, to exploring stance trends in a corpus through an interactive dashboard. To support this, we provide stance target extraction, stance detection, stance trend inference, and an exploratory dashboard, all available in an easy-to-use library. We hope that this library can increase the accessibility of stance detection for the wider community of those who could benefit from this method.

StanceMining: An open-source stance detection library supporting time-series and visualization

Short-form video platforms like TikTok present unique challenges for misinformation detection due to their multimodal, dynamic, and noisy content. We present ShortCheck, a modular, inference-only pipeline with a user-friendly pipeline that automatically identifies checkworthy short-form videos to help human fact-checkers. The system integrates speech transcription, OCR, object and deepfake detection, video-to-text summarization, and claim verification. ShortCheck is validated by evaluating it on two manually annotated datasets with TikTok videos in a multilingual setting. The pipeline achieves promising results with F1-weighted score over 70%. The demo can be accessed live at \url{http://shortcheck.factiverse.ai}.

ShortCheck: Checkworthiness Detection of Multilingual Short‑Form Videos

Accurate assessment of generated chart quality is crucial for automated document creation and editing across diverse applications like finance, medicine, policy making, and education. Current evaluation approaches suffer from significant limitations: human evaluation is costly and difficult to scale, pixel-based metrics ignore data accuracy, while data-centric measures overlook design quality. Recent multimodal LLM evaluators show promise but exhibit concerning inconsistencies due to prompt sensitivity and subjective biases. Existing metrics fail to evaluate chart quality holistically across visual similarity, semantic alignment, and data fidelity, often producing misleading scores that unfairly penalize good charts while rewarding bad ones. We introduce ChartEval, a novel chart evaluation system that compares generated chart images with ground truth by leveraging scene graph parsing to decompose chart images into hierarchical scene graphs of chart objects, attributes, and relations. Subsequently, it applies graph-based similarity measures to compare candidate chart scene graphs against reference scene graphs for measuring chart quality. We demonstrate that our evaluation approach achieves significantly stronger correlation with human judgments compared to existing metrics like GPT-Score, SSIM, and SCRM using a comprehensive benchmark of 4K chart images paired with generation intents and human quality ratings. We demonstrate the utility of the ChartEval system as a reliable automatic chart quality metric on diverse tasks, including language-guided chart editing, chart reconstruction, and text-to-chart synthesis using both open-source and API-based LLMs.

Downloads

Next from IJCNLP-AACL 2025

Noise May Drown Out Words but Foster Compositionality: The Advantage of the Erasure and Deletion Noisy Channels on Emergent Communication

Similar lecture

Benchmarking Systematic Relational Reasoning with Large Language and Reasoning Models