China

Recent advancements in language models have led to significant improvements in mathematical reasoning across various benchmarks. However, most of these benchmarks rely on automatic evaluation methods that only compare final answers using heuristics, without verifying the underlying reasoning steps. This limitation results in false positive solutions, where models may produce correct final answers but with flawed deduction paths. In this paper, we systematically examine the prevalence of false positive solutions in mathematical problem solving for language models. We analyze the characteristics and extent of this issue across different open-source models, datasets of varying difficulty levels, and decoding strategies. Specifically, we explore how false positives influence the inference time scaling behavior of language models. Our experimental results reveal that: (1) false positive solutions persist across different models, datasets, and decoding methods, (2) sampling-based inference time scaling methods do not alleviate the problem, and (3) the pass@N evaluation metric is more susceptible to false positives, suggesting a significantly lower scaling ceiling than what automatic evaluations indicate. Additionally, we analyze specific instances of false positives and discuss potential limitations in self-improvement techniques and synthetic data generation under such conditions.

EMNLP 2025

Examining False Positives under Inference Scaling for Mathematical Reasoning

inference methods

human evaluation

automatic evaluation

analysis

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields 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.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. 
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/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://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</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 ACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.aclweb.org/registration/) first.

Registration Required

ACL 2026

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

poster

## Welcome!
"I am excited to welcome you to this year’s edition of the Conference on Empirical Methods in Natural Language Processing! Importantly, it marks the 30th edition of EMNLP. With over 8,000 submissions, more than 3,000 accepted papers, and thousands of attendees, we have come a long way from that first
workshop, which had 14 accepted papers. As the field looks ahead, Suzhou is the fitting location for celebrating this milestone: rooted in a long literary tradition, yet modern and forward-looking, and home to a large share of the NLP community." 

*Message from the General Chair, Dirk Hovy*

Celebrate 30 Years of EMNLP! 
EMNLP 2025 will be held in Suzhou, China from November 5th to November 9th, 2025.

We introduce ProcWorld, a large-scale benchmark for partially observable embodied spatial reasoning and long-term planning with large language models (LLM) and vision language models (VLM). ProcWorld features a wide range of challenging embodied navigation and object manipulation tasks, covering 16 task types, 5,000 rooms, and over 10 million evaluation trajectories with diverse data distribution. ProcWorld supports configurable observation modes, ranging from text-only descriptions to vision-only observations. It enables text-based actions to control the agent following language instructions. ProcWorld has presented significant challenges for LLMs and VLMs: (1) active information gathering given partial observations for disambiguation; (2) simultaneous localization and decision-making by tracking the spatio-temporal state-action distribution; (3) constrained reasoning with dynamic states subject to physical reachability. Our extensive evaluation of 15 foundation models and 5 reasoning algorithms (with over 1 million rollouts) indicates larger models perform better. However, ProcWorld remains highly challenging for existing state-of-the-art models and in-context learning methods due to constrained reachability and the need for combinatorial spatial reasoning.

ProcWorld: Benchmarking Large Model Planning in Reachability-Constrained Environments

Reasoning is a fundamental capability often required in real-world text-to-image (T2I) generation, e.g., generating "a bitten apple that has been left in the air for more than a week" necessitates understanding temporal decay and commonsense concepts. While recent T2I models have made impressive progress in producing photorealistic images, their reasoning capability remains underdeveloped and insufficiently evaluated. To bridge this gap, we introduce R2I-Bench, a comprehensive benchmark specifically designed to rigorously assess reasoning-driven T2I generation. R2I-Bench comprises 3068 meticulously curated data instances, spanning 7 core reasoning categories, including commonsense, mathematical, logical, compositional, numerical, causal, and concept mixing. To facilitate fine-grained evaluation, we design R2IScore, a QA-style metric based on instance-specific, reasoning-oriented evaluation questions that assess three critical dimensions: text-image alignment, reasoning accuracy, and image quality. Extensive experiments with 16 representative T2I models, including a strong pipeline-based framework that decouples reasoning and generation using the state-of-the-art language and image generation models, demonstrate consistently limited reasoning performance, highlighting the need for more robust, reasoning-aware architectures in the next generation of T2I systems.

R2I-Bench: Benchmarking Reasoning-Driven Text-to-Image Generation

Existing table reasoning methods face challenges due to complex table structures and intricate logical reasoning. While supervised fine-tuning (SFT) dominates existing research, reinforcement learning (RL, such as Group Relative Policy Optimization: GRPO) has shown promise but struggled with low initial policy accuracy and coarse rewards in tabular contexts. In this paper, we introduce Table-R1, a three-stage RL framework that enhances multimodal table reasoning through: (1) Warm-up that prompts initial perception and reasoning capabilities, (2) Perception Alignment GRPO (PA-GRPO), which employs continuous Tree-Edit-Distance Similarity (TEDS) rewards for recognizing table structures and contents, and (3) Hint-Completion GRPO (HC-GRPO), which utilizes fine-grained rewards of residual steps based on the hint-guided question. Extensive experiments demonstrate that Table-R1 can boost the model's table reasoning performance obviously on both held-in and held-out datasets, outperforming SFT and GRPO largely. Notably, the Qwen2-VL-7B model with Table-R1 can surpass larger specific tabular reasoning open-source Large Vision Language Models (e.g., 13B Table-LLaVa), even achieving comparable performance to the closed-source model GPT-4o on held-in datasets. Our work highlights the efficacy of each stage of Table-R1 in overcoming initialization bottlenecks and reward sparsity, thereby boosting the GRPO on robust multimodal table reasoning for real-world applications.

Can GRPO Boost Complex Multimodal Table Understanding?

Large Language Models (LLMs) demonstrate remarkable capabilities in text understanding and generation. However, their tendency to produce factually inconsistent outputs—commonly referred to as ``hallucinations''—remains a critical challenge. Existing approaches, such as retrieval-based and inference-time correction methods, primarily address this issue at the input or output level, often overlooking the intrinsic information refinement process and the role of premature layers. Meanwhile, alignment- and fine-tuning-based methods are resource-intensive. In this paper, we propose **PLI** (**P**remature **L**ayers **I**nterpolation), a novel, training-free, and plug-and-play intervention designed to enhance factuality. PLI mitigates hallucinations by inserting premature layers formed through mathematical interpolation with adjacent layers. Inspired by stable diffusion and sampling steps, PLI extends the depth of information processing and transmission in LLMs, improving factual coherence. Experiments on four publicly available datasets demonstrate that PLI effectively reduces hallucinations while outperforming existing baselines in most cases. Further analysis suggests that the success of layer interpolation is closely linked to LLMs' internal mechanisms. To promote reproducibility, we will release our code and data upon acceptance.

Expanding before Inferring: Enhancing Factuality in Large Language Models through Premature Layers Interpolation

Panic attacks are acute episodes of fear and distress, in which timely, appropriate intervention can significantly help individuals regain stability. However, suitable datasets for training such models remain scarce due to ethical and logistical issues. To address this, we introduce Pace, which is a dataset that includes high-distress episodes constructed from first-person narratives, and structured around the principles of Psychological First Aid (PFA). Using this data, we train Pacer, a counseling model designed to provide both empathetic and directive support, which is optimized through supervised learning and simulated preference alignment. To assess its effectiveness, we propose PanicEval, a multi-dimensional framework covering general counseling quality and crisis-specific strategies. Experimental results show that Pacer outperforms strong baselines in both counselor-side metrics and client affect improvement. Human evaluations further confirm its practical value, with Pacer consistently preferred over general, CBT-based, and GPT-4-powered models in panic scenarios.

PanicToCalm: A Proactive Counseling Agent for Panic Attacks

Recent advancements in music large language models (LLMs) have significantly improved music understanding tasks, which involve the model’s ability to analyze and interpret various musical elements. These improvements primarily focused on integrating both music and text inputs. However, the potential of incorporating additional modalities such as images, videos and textual music features to enhance music understanding remains unexplored. To bridge this gap, we propose DeepResonance, a multimodal music understanding LLM fine-tuned via multi-way instruction tuning with multi-way aligned music, text, image, and video data. To this end, we construct Music4way-MI2T, Music4way-MV2T, and Music4way-Any2T, three 4-way training and evaluation datasets designed to enable DeepResonance to integrate both visual and textual music feature content. We also introduce multi-sampled ImageBind embeddings and a pre-LLM fusion Transformer to enhance modality fusion prior to input into text LLMs, tailoring DeepResonance for multi-way instruction tuning. Our model achieves state-of-the-art performances across six music understanding tasks, highlighting the benefits of the auxiliary modalities and the structural superiority of DeepResonance. We plan to open-source the models and the newly constructed datasets.

DeepResonance: Enhancing Multimodal Music Understanding via Music-centric Multi-way Instruction Tuning

Symbolic regression is a powerful technique for discovering mathematical expressions that best fit observed data. While neural symbolic regression methods based on large-scale pre-trained models perform well on simple tasks, the reliance on fixed parametric knowledge typically limits their generalization to complex and diverse data distributions. To address this challenge, we propose a syntax-aware retrieval-augmented mechanism that leverages the syntactic structure of symbolic expressions to perform context-aware retrieval from a pre-constructed token datastore during inference. This mechanism enables the model to incorporate highly relevant non-parametric prior information to assist in expression generation. Additionally, we design an entropy-based confidence network that dynamically adjusts the fusion strength between neural and retrieved components by estimating predictive uncertainty. Extensive experiments on multiple symbolic regression benchmarks demonstrate that the proposed method significantly outperforms representative baselines, validating the effectiveness of retrieval augmentation in enhancing the generalization performance of neural symbolic regression models.

Syntax-Aware Retrieval Augmentation for Neural Symbolic Regression

Recent advances in Multimodal Large Language Models (MLLMs) have enhanced their versatility as they integrate a growing number of modalities. Considering the heavy cost of training MLLMs, it is efficient to reuse the existing ones and extend them to more modalities through Modality-incremental Continual Learning (MCL). The exploration of MCL is in its early stages. In this work, we dive into the causes of performance degradation in MCL. We uncover that it suffers not only from forgetting as in traditional continual learning, but also from misalignment between the modality-agnostic and modality-specific components. To this end, we propose an elegantly simple MCL paradigm called "MErge then ReAlign" (MERA) to address both forgetting and misalignment. MERA avoids introducing heavy model budgets or modifying model architectures, hence is easy to deploy and highly reusable in the MLLM community. Extensive experiments demonstrate the impressive performance of MERA, holding an average of 99.84% Backward Relative Gain when extending to four modalities, achieving nearly lossless MCL performance. Our findings underscore the misalignment issue in MCL. More broadly, our work showcases how to adjust different components of MLLMs during continual learning.

Merge then Realign: Simple and Effective Modality-Incremental Continual Learning for Multimodal LLMs

Vision-language-action models have gained significant attention for their ability to model multimodal sequences in embodied instruction following tasks. However, most existing models rely on causal attention, which we find to be suboptimal for processing sequences composed of interleaved segments from different modalities. In this paper, we introduce Actra, a novel Transformer architecture featuring trajectory attention and learnable action queries, designed to efficiently process segmented multimodal trajectories and predict 6D pose actions for imitation learning. Furthermore, we propose a contrastive dynamics learning objective to enhance the model's understanding of environment dynamics and multimodal alignment, complementing the primary behavior cloning objective. Through extensive experiments on three large-scale robot manipulation benchmarks, Actra demonstrates substantial performance improvements over previous models.

Astra: Efficient Transformer Architecture and Contrastive Dynamics Learning for Embodied Instruction Following

Hate speech detection is key to online content moderation, but current models struggle to generalise beyond their training data. This has been linked to dataset biases and the use of sentence-level labels, which fail to teach models the underlying structure of hate speech. In this work, we show that even when models are trained with more fine-grained, span-level annotations (e.g., "artists" is labeled as target and "are parasites" as dehumanising comparison), they struggle to disentangle the meaning of these labels from the surrounding context. As a result, combinations of expressions that deviate from those seen during training remain particularly difficult for models to detect. We investigate whether training on a dataset where expressions occur with equal frequency across all contexts can improve generalisation. To this end, we create U-PLEAD, a dataset of ~364,000 synthetic posts, along with a novel compositional generalisation benchmark of ~8,000 manually validated posts. Training on a combination of U-PLEAD and real data improves compositional generalisation while achieving state-of-the-art performance on the human-sourced PLEAD.

Downloads

Next from EMNLP 2025

ProcWorld: Benchmarking Large Model Planning in Reachability-Constrained Environments

Similar lecture

ARISE: An Adaptive Resolution-Aware Metric for Test-Time Scaling Evaluation in Large Reasoning Models