China

We introduce Refeeding State Embeddings aligned using Environmental Data (ReSEED), a novel method for grounding language in environmental data. While large language models (LLMs) excel at many tasks, they continue to struggle with multi-step sequential reasoning. ReSEED addresses this by producing latent embeddings aligned with the true state of the environment and refeeding these embeddings into the model before generating its output. To evaluate its effectiveness, we develop three new sequential reasoning benchmarks, each with a training set of paired state-text trajectories and several text-only evaluation sets that test generalization to longer, unseen trajectories. Across all benchmarks, ReSEED significantly improves generalization and scalability over a text-only baseline. We further show that ReSEED outperforms commercial LLMs on our benchmarks, highlighting the value of grounding language in the environment.

EMNLP 2025

ReSeeding Latent States for Sequential Language Understanding

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."<br>

*Message from the General Chair, Dirk Hovy*

[**Link to Conference Handbook**](https://drive.google.com/file/d/1johU5QqVVYO4RfH7QcIORr7qrVBdzdwC/view?usp=sharing)





<br>

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

We introduce CSAR, an algorithm for inducing morphemes from emergent language corpora of parallel utterances and meanings. It is a greedy algorithm that (1) weights morphemes based on mutual information between forms and meanings, (2) selects the highest-weighted pair, (3) removes it from the corpus, and (4) repeats the process to induce further morphemes (i.e., Count, Select, Ablate, Repeat). The effectiveness of CSAR is first validated on procedurally generated datasets and compared against baselines for related tasks. Second, we validate CSAR's performance on human language data to show that the algorithm makes reasonable predictions in adjacent domains. Finally, we analyze a handful of emergent languages, quantifying linguistic characteristics like degree of synonymy and polysemy.

Morpheme Induction for Emergent Language

Data quality and its effective selection are fundamental to improving the performance of machine translation models, serving as cornerstones for achieving robust and reliable translation systems. This paper presents a data selection methodology specifically designed for fine-tuning machine translation systems, which leverages the synergy between a learner model and a pre-trained reference model to enhance overall training effectiveness. By defining a learnability score, our approach systematically evaluates the utility of data points for training, ensuring that only the most relevant and impactful examples contribute to the fine-tuning process. Furthermore, our method employs a batch selection strategy which considers interdependencies among data points, optimizing the efficiency of the training process while maintaining a focus on data relevance. Experiments on English ↔ Persian and several other language pairs using an mBART model fine-tuned on the CCMatrix dataset demonstrate that our method can achieve up to a fivefold improvement in data efficiency compared to an iid baseline. Experimental results indicate that our approach improves computational efficiency by 24 when utilizing cached embeddings, as it requires fewer training data points. Additionally, it enhances generalization, resulting in superior translation performance compared to random selection method.

Dynamic Jointly Batch Selection for Data Efficient Machine Translation Fine-Tuning

Large language models (LLMs) often leverage adapters, such as low-rank-based adapters, to achieve strong performance on downstream tasks. However, storing a separate adapter for each task significantly increases memory requirements, posing a challenge for resource-constrained environ ments such as mobile devices. Although model merging techniques can reduce storage costs, they typically result in substantial performance degradation. In this work, we introduce HydraOpt, a new model merging technique that capitalizes on the inherent similarities between the matrices of low-rank adapters. Unlike existing methods that produce a fixed trade-off between storage size and performance, HydraOpt allows us to navigate this spectrum of efficiency and performance. Our experiments show that HydraOpt significantly reduces storage size (48% reduction) compared to storing all adapters, while achieving competitive performance (0.2-1.8% drop). Furthermore, it outperforms existing merging techniques in terms of performance at the same or slightly worse storage efficiency.

HydraOpt: Navigating the Efficiency-Performance Trade-off of Adapter Merging

Scientific databases aggregate vast amounts of quantitative data alongside descriptive text. In biochemistry, chemical screening assays evaluate the functional responses of candidate compounds against disease targets. Unstructured text that describes the biological mechanisms through which these targets operate, experimental screening protocols, and other attributes of assays offer rich information for new drug discovery campaigns, but has been untapped because of that unstructured format. We present Assay2Mol, a large language model-based workflow that can capitalize on the vast existing biochemical screening assays for early-stage drug discovery. Assay2Mol retrieves existing assay records involving targets similar to the new target and generates candidate compounds using in-context learning with the retrieved assay screening data. Assay2Mol outperforms recent machine learning approaches that generate candidate ligand compounds for target protein structures, while also promoting more synthesizable molecule generation.

Assay2Mol: Large Language Model-based Drug Design Using BioAssay Context

Social and behavioral determinants of health (SBDH) play a crucial role in health outcomes and are frequently documented in clinical text. Automatically extracting SBDH information from clinical text relies on publicly available good-quality datasets. However, existing SBDH datasets exhibit substantial limitations in their availability and coverage. In this study, we introduce Synth-SBDH, a novel synthetic dataset with detailed SBDH annotations, encompassing status, temporal information, and rationale across 15 SBDH categories. We showcase the utility of Synth-SBDH on three tasks using real-world clinical datasets from two distinct hospital settings, highlighting its versatility, generalizability, and distillation capabilities. Models trained on Synth-SBDH consistently outperform counterparts with no Synth-SBDH training, achieving up to 63.75% macro-F improvements. Additionally, Synth-SBDH proves effective for rare SBDH categories and under-resource constraints while being substantially cheaper than expert-annotated real-world data. Human evaluation reveals a 71.06% Human-LLM alignment and uncovers areas for future refinements.

Synth-SBDH: A Synthetic Dataset of Social and Behavioral Determinants of Health for Clinical Text

Large Language Models (LLMs) offer natural language explanations as an alternative to feature attribution methods for model interpretability. However, despite their plausibility, they may not reflect the model's internal reasoning faithfully, which is crucial for understanding the model's true decision-making processes. Although several faithfulness metrics have been proposed, a unified evaluation framework remains absent. Here, we present Causal Diagnosticity, a framework to evaluate faithfulness metrics for natural language explanations. Our framework employs the concept of diagnosticity, and uses model-editing methods to generate faithful-unfaithful explanation pairs. Our benchmark includes four tasks: fact-checking, analogy, object counting, and multi-hop reasoning. We evaluate a prominent faithfulness metrics, including post-hoc explanation and chain-of-thought-based methods. We find that diagnostic performance varies across tasks and models, with Filler Tokens performing best overall. Additionally, continuous metrics are generally more diagnostic than binary ones but can be sensitive to noise and model choice. Our results highlight the need for more robust faithfulness metrics.

A Causal Lens for Evaluating Faithfulness Metrics

Societal stereotypes are at the center of a myriad of responsible AI interventions targeted at reducing the generation and propagation of potentially harmful outcomes. While these efforts are much needed, they tend to be fragmented and often address different parts of the issue without taking in a unified or holistic approach about social stereotypes and how they impact various parts of the machine learning pipeline. As a result, it fails to capitalize on the underlying mechanisms that are common across different types of stereotypes, and to anchor on particular aspects that are relevant in certain cases. In this paper, we draw on social psychological research, and build on NLP data and methods, to propose a unified framework to operationalize stereotypes in generative AI evaluations. Our framework identifies key components of stereotypes that are crucial in AI evaluation, including the target group, associated attribute, relationship characteristics, perceiving group, and relevant context. We also provide considerations and recommendations for its responsible use.

A Comprehensive Framework to Operationalize Social Stereotypes for Responsible AI Evaluations

Large language models (LLMs) are increasingly deployed in domains requiring moral understanding, yet their reasoning often remains shallow and misaligned with human reasoning. Unlike humans, whose moral reasoning integrates contextual trade-offs, value systems, and ethical theories, LLMs often rely on surface patterns, leading to biased decisions in morally and ethically complex scenarios. To address this gap, we present a value-grounded framework for evaluating and distilling structured moral reasoning in LLMs. We benchmark 12 open-source models across four moral datasets using a taxonomy of prompts grounded in value systems, ethical theories, and cognitive reasoning strategies. Our evaluation is guided by four questions: (1) Does reasoning improve LLM decision-making over direct prompting? (2) Which types of value/ethical frameworks most effectively guide LLM reasoning? (3) Which cognitive reasoning strategies lead to better moral performance? (4) Can small-sized LLMs acquire moral competence through distillation? We find that prompting with explicit moral structure consistently improves accuracy and coherence, with first-principles reasoning and Schwartz's + care-ethics scaffolds yielding the strongest gains. Furthermore, our supervised distillation approach transfers moral competence from large to small models without additional inference cost. Together, our results offer a scalable path toward interpretable and value-grounded models.

Structured Moral Reasoning in Language Models: A Value-Grounded Evaluation Framework

We present Cacheback Decoding, a training-free model-agnostic speculative decoding method. Cacheback leverages only a Least Recently Used (LRU) cache of token n-grams to generate draft sequences. Despite its minimalist design, it achieves state-of-the-art performance among comparable methods.

Cacheback: Speculative Decoding With Nothing But Cache

Large language model (LLM) reasoning can be improved by scaling test-time compute with aggregation, i.e., generating multiple samples and aggregating over them. While improving performance, this strategy often reaches a saturation point beyond which additional compute provides no return. Refinement offers an alternative by using model-generated feedback to improve answer quality. However, refinement faces three key challenges: (1) Excessive refinement: Uniformly refining all instances can cause over-correction and reduce overall performance. (2) Inability to localize and address errors: LLMs struggle to identify and correct their own mistakes. (3) Insufficient refinement: Stopping refinement too soon could leave errors unaddressed. To tackle these issues, we propose MAgICoRe, a framework for Multi-Agent Iteration for Coarse-to-fine Refinement. MAgICoRe mitigates excessive refinement by categorizing problems as easy or hard, solving easy problems with coarse-grained aggregation, and solving the hard ones with fine-grained multi-agent refinement. To better localize errors, we incorporate external step-wise reward model scores, and to ensure sufficient refinement, we iteratively refine the solutions using a multi-agent setup. We evaluate MAgICoRe on Llama-3-8B and GPT- 3.5 and show its effectiveness across seven reasoning datasets. One iteration of MAgICoRe beats Self-Consistency by 3.4%, Best-of-k by 3.2%, and Self-Refine by 4.0% even when these baselines use k = 120, and MAgICoRe uses less than 50% of the compute.

Downloads

Next from EMNLP 2025

Morpheme Induction for Emergent Language