Austria

Text embeddings from large language models (LLMs) have achieved excellent results in tasks such as information retrieval, semantic textual similarity, etc. In this work, we show an interesting finding: when feeding a text into the LLM-based embedder, the obtained text embedding will be able to be aligned with the key tokens in the input text. We first fully analyze this phenomenon on eight LLM-based embedders and show that this phenomenon is universal and is not affected by model architecture, training strategy, and embedding method. With a deeper analysis, we find that the main change in embedding space between these embedders and their LLM backbones is in the first principal component. By adjusting the first principal component, we can align text embedding with the key tokens. Finally, we give several examples to demonstrate the vast application potential of this finding: (1) we propose a simple and practical sparse retrieval method based on the aligned tokens, which can achieve 80% of the dense retrieval effect of the same model while reducing the computation significantly; (2) we show that our findings provide a novel perspective to help understand novel technologies (e.g., instruction-following embedding) and fuzzy concepts (e.g., semantic relatedness vs. similarity) in this field.

ACL 2025

A Text is Worth Several Tokens: Text Embedding from LLMs Secretly Aligns Well with The Key Tokens

intruction-following embedding

text embedding

large language model

sparse retrieval

## 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*

EMNLP 2025

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

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)

Evidence-enhanced detectors present remarkable abilities in identifying malicious social text. However, the rise of large language models (LLMs) brings potential risks of evidence pollution to confuse detectors. This paper explores potential manipulation scenarios including basic pollution, and rephrasing or generating evidence by LLMs. To mitigate the negative impact, we propose three defense strategies from the data and model sides, including machine-generated text detection, a mixture of experts, and parameter updating. Extensive experiments on four malicious social text detection tasks with ten datasets illustrate that evidence pollution significantly compromises detectors, where the generating strategy causes up to a 14.4% performance drop. Meanwhile, the defense strategies could mitigate evidence pollution, but they faced limitations for practical employment. Further analysis illustrates that polluted evidence (i) is of high quality, evaluated by metrics and humans; (ii) would compromise the model calibration, increasing expected calibration error up to 21.6%; and (iii) could be integrated to amplify the negative impact, especially for encoder-based LMs, where the accuracy drops by 21.8%.

On the Risk of Evidence Pollution for Malicious Social Text Detection in the Era of LLMs

Historical analogies, which compare known past events with contemporary but unfamiliar events, are important abilities that help people make decisions and understand the world. However, research in applied history suggests that people have difficulty finding appropriate analogies. And previous studies in the AI community have also overlooked historical analogies. To fill this gap, in this paper, we focus on the historical analogy acquisition task, which aims to acquire analogous historical events for a given event. We explore retrieval and generation methods for acquiring historical analogies based on different large language models (LLMs). Furthermore, we propose a self-reflection method to mitigate hallucinations and stereotypes when LLMs generate historical analogies. Through human evaluations and our specially designed automatic multi-dimensional assessment, we find that LLMs generally have a good potential for historical analogies. And the performance of the models can be further improved by using our self-reflection method. Resources of this paper can be found at https://anonymous.4open.science/r/Historical-Analogy-of-LLMs-FC17

Past Meets Present: Creating Historical Analogy with Large Language Models

Recent advancements in visual generative models have enabled high-quality image and video generation, opening diverse applications. However, evaluating these models often demands sampling hundreds or thousands of images or videos, making the process computationally expensive, especially for diffusion-based models with inherently slow sampling. Moreover, existing evaluation methods rely on rigid pipelines that overlook specific user needs and provide numerical results without clear explanations. In contrast, humans can quickly form impressions of a model's capabilities by observing only a few samples. To mimic this, we propose the Evaluation Agent framework, which employs human-like strategies for efficient, dynamic, multi-round evaluations using only a few samples per round, while offering detailed, user-tailored analyses. It offers four key advantages: 1) efficiency, 2) promptable evaluation tailored to diverse user needs, 3) explainability beyond single numerical scores, and 4) scalability across various models and tools. Experiments show that Evaluation Agent reduces evaluation time to 10% of traditional methods while delivering comparable results. The Evaluation Agent framework is fully open-sourced to advance research in visual generative models and their efficient evaluation.

Evaluation Agent: Efficient and Promptable Evaluation Framework for Visual Generative Models

Multimodel Large Language Models(MLLMs) have achieved promising OCR-free Document Understanding performance by increasing the supported resolution of document images. However, this comes at the cost of generating thousands of visual tokens for a single document image, leading to excessive GPU memory and slower inference times, particularly in multi-page document comprehension. In this work, to address these challenges, we propose a High-resolution DocCompressor module to compress each high-resolution document image into 324 tokens, guided by low-resolution global visual features. With this compression module, to strengthen multi-page document comprehension ability and balance both token efficiency and question-answering performance, we develop the DocOwl2 under a three-stage training framework: Single-image Pretraining, Multi-image Continue-pretraining, and Multi-task Finetuning. DocOwl2 sets a new state-of-the-art across multi-page document understanding benchmarks and reduces first token latency by more than 50%. Compared to single-image MLLMs trained on similar data, our DocOwl2 achieves comparable single-page understanding performance with less than 20% of the visual tokens. Our codes, models, and data will be publicly available.

mPLUG-DocOwl2: High-resolution Compressing for OCR-free Multi-page Document Understanding

While long-context inference is crucial for advancing large language model (LLM) applications, its prefill speed remains a significant bottleneck. Current approaches, including sequence parallelism strategies and compute reduction through approximate attention mechanisms, still fall short of delivering optimal inference efficiency. This hinders scaling the inputs to longer sequences and processing long-context queries in a timely manner. To address this, we introduce APB, an efficient long-context inference framework that leverages multi-host approximate attention to enhance prefill speed by reducing compute and enhancing parallelism simultaneously. APB introduces a communication mechanism for essential key-value pairs within a sequence parallelism framework, enabling a faster inference speed while maintaining task performance. We implement APB by incorporating a tailored FlashAttn kernel alongside optimized distribution strategies, supporting diverse models and parallelism configurations. APB achieves speedups of up to 9.2$\times$, 4.2$\times$, and 1.6$\times$ compared with FlashAttn, RingAttn, and StarAttn, respectively, without any observable task performance degradation.

APB: Accelerating Distributed Long-Context Inference by Passing Compressed Context Blocks across GPUs

Ensuring awareness of fairness and privacy in Large Language Models (LLMs) is critical. Interestingly, we discover a counter-intuitive trade-off phenomenon that enhancing an LLM's privacy awareness through Supervised Fine-Tuning (SFT) methods significantly decreases its fairness awareness with thousands of samples. To address this issue, inspired by the information theory, we introduce a training-free method to \textbf{S}uppress the \textbf{P}rivacy and fa\textbf{I}rness coupled \textbf{N}eurons (\textbf{SPIN}), which theoretically and empirically decrease the mutual information between fairness and privacy awareness. Extensive experimental results demonstrate that SPIN eliminates the trade-off phenomenon and significantly improves LLMs' fairness and privacy awareness simultaneously without compromising general capabilities, e.g., improving Qwen-2-7B-Instruct's fairness awareness by 12.2\% and privacy awareness by 14.0\%.
More crucially, SPIN remains robust and effective with limited annotated data or even when only malicious fine-tuning data is available, whereas SFT methods may fail to perform properly in such scenarios. 
Furthermore, we show that SPIN could generalize to other potential trade-off dimensions.
We hope this study provides valuable insights into concurrently addressing fairness and privacy concerns in LLMs and can be integrated into comprehensive frameworks to develop more ethical and responsible AI systems. Our code is available at \url{https://github.com/ChnQ/SPIN}.

The Tug of War Within: Mitigating the Fairness-Privacy Conflicts in Large Language Models

While retrieval techniques are widely used in practice, they still face significant challenges in cross-domain scenarios. Recently, generation-augmented methods have emerged as a promising solution to this problem. These methods enhance raw queries by incorporating additional information from an LLM-based generator, facilitating more direct retrieval of relevant documents. However, existing methods struggle with highly specialized situations that require extensive domain expertise. To address this problem, we present \textbf{Reinforced-IR}, a novel approach that jointly adapts a pre-trained retriever and generator for precise cross-domain retrieval. A key innovation of Reinforced-IR is its \textbf{Self-Boosting} framework, which enables retriever and generator to learn from each other's feedback. Specifically, the generator is reinforced to generate query augmentations that enhance the retriever's performance, while the retriever is trained to better discriminate the relevant documents identified by the generator. This iterative process allows the end-to-end retrieval performance to be progressively optimized using an unlabeled corpus from the target domain. In our experiment, Reinforced-IR outperforms existing domain adaptation methods by a large margin, leading to substantial improvements in retrieval quality across a wide range of application scenarios.
We have publicly released our code at this \href{https://github.com/FlagOpen/FlagEmbedding/tree/master/research/Reinforced_IR}{repo}.

Reinforced IR: A Dual Reinforcement Framework For Domain-Adapted Information Retrieval

Existing approaches to fine-grained emotion classification (FEC) often operate in Euclidean space, where the flat geometry limits the ability to distinguish semantically similar emotion labels (e.g., *annoyed* vs. *angry*). While prior research has explored hyperbolic geometry to capture fine-grained label distinctions, it typically relies on predefined hierarchies and overlooks semantically confusable negatives. In this work, we propose HyCoEM, a semantic alignment framework that leverages the Lorentz model of hyperbolic space. Our approach jointly embeds text and label representations into hyperbolic space via the exponential map, and employs a contrastive loss to bring text embeddings closer to their true labels while pushing them away from adaptively selected, semantically similar negatives. This enables the model to learn label embeddings without relying on a predefined hierarchy and better captures subtle distinctions by incorporating information from both positive and challenging negative labels. Experimental results on two benchmark FEC datasets demonstrate the effectiveness of our approach over baseline methods.

Semantic alignment in hyperbolic space for fine-grained emotion classification

Food delivery search aims to quickly retrieve deliverable items that meet users' needs, typically requiring faster and more accurate query understanding compared to traditional e-commerce search. Generative retrieval (GR), an emerging search paradigm, harnesses the advanced query understanding capabilities of large language models (LLMs) to enhance the retrieval of results for complex and long-tail queries in food delivery search scenarios. However, there are still challenges in deploying GR to online scenarios: 1) **the large scale of items**; 2) **latency constraints unmet by LLM inference in online retrieval**; and 3) **strong location-based service restrictions on generated items**. To explore the application of GR in food delivery search, we optimize both offline training and online deployment, proposing **Hier**archical semantic representation enhancement for **G**enerative **R**etrieval (HierGR). Specifically, for the generation of semantic IDs, we propose an optimization method that refines the residual quantization process to generate hierarchically semantic IDs for items. Additionally, to successfully deploy on a well-known food delivery platform, we utilize the query cache mechanism and integrate the GR model with the online dense retrieval model to fulfill real-world search requirements. Online A/B testing results show that our proposed method increases **the number of online orders by 0.68\%** for complex search intents. The source code is available at https://github.com/zhangfw123/HierGR.

HierGR: Hierarchical Semantic Representation Enhancement for Generative Retrieval in Food Delivery Search

Multilingual language models (MLLMs) have demonstrated remarkable abilities to transfer knowledge across languages, despite being trained without explicit cross-lingual supervision. We analyze the parameter spaces of three MLLMs to study how their representations evolve during pre-training, observing patterns consistent with compression: models initially form language-specific representations, which gradually converge into cross-lingual abstractions as training progresses. Through probing experiments, we observe a clear transition from uniform language identification capabilities across layers to more specialized layer functions. For deeper analysis, we focus on neurons that encode distinct semantic concepts. By tracing their development during pre-training, we show how they gradually align across languages. Notably, we identify specific neurons that emerge as increasingly reliable predictors for the same concepts across languages. This alignment manifests concretely in generation: once an MLLM exhibits cross-lingual generalization according to our measures, we can select concept-specific neurons identified from, e.g., Spanish text and manipulate them to guide token predictions. Remarkably, rather than generating Spanish text, the model produces semantically coherent English text. This demonstrates that cross-lingually aligned neurons encode generalized semantic representations, independent of the original language encoding.

Downloads

Next from ACL 2025

On the Risk of Evidence Pollution for Malicious Social Text Detection in the Era of LLMs

Similar lecture

Diffusion vs. Autoregressive Language Models: A Text Embedding Perspective