United States

Starting from the hypothesis that knowledge in semantic space is organized along structured manifolds, we argue that this geometric structure renders the space explorable. By traversing it and using the resulting continuous representations to condition an LLM’s generation distribution, we can systematically expand the model’s reachable semantic range. We introduce a framework that requires no modification of LLM parameters and operationalizes this idea by constructing a conditioning distribution from a small set of diverse anchor generations. This distribution conditions LLM&#39;s generation via an xRAG-style projector. Our experiments demonstrate that this manifold-based conditioning substantially increases generative diversity, with direct benefits for enhancing divergent thinking, a core facet of creativity, in language models.

ACL 2026

 Geometry of Knowledge Allows Extending Diversity Boundaries of Large Language Models

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. <br>
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) <br><br><br>
<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>
<br><br><br><br>

To access this event page, you need to log in with the **email address you registered with**. <br>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

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

Starting from the hypothesis that knowledge in semantic space is organized along structured manifolds, we argue that this geometric structure renders the space explorable. By traversing it and using the resulting continuous representations to condition an LLM’s generation distribution, we can systematically expand the model’s reachable semantic range. We introduce a framework that requires no modification of LLM parameters and operationalizes this idea by constructing a conditioning distribution from a small set of diverse anchor generations. This distribution conditions LLM's generation via an xRAG-style projector. Our experiments demonstrate that this manifold-based conditioning substantially increases generative diversity, with direct benefits for enhancing divergent thinking, a core facet of creativity, in language models.

poster

We propose a lightweight and single-pass uncertainty quantification method for detecting hallucinations in Large Language Models. The method uses attention matrices to estimate uncertainty without requiring repeated sampling or external models. Specifically, we measure the Kullback–Leibler divergence between each attention head’s distribution and a uniform reference distribution, and use these features in a logistic regression probe. Across multiple datasets, task types, and model families, attention divergence is strongly predictive of answer correctness and performs competitively with existing uncertainty estimation methods. We find that this signal is concentrated in middle layers and on factual tokens such as named entities and numbers, suggesting that attention dynamics provides an efficient and interpretable white-box signal of model uncertainty.

 Detecting Hallucinations in Large Language Models via Internal Attention Divergence Signals

Retrieval-augmented generation is often bottlenecked by retrieval quality, motivating query expansion. We propose RAQE (Reranker-Aligned Query Expansion), a label-free framework that trains a compact generator (3B/8B) via policy optimization using a reranker-derived reward. RAQE replaces relevance labels with Reranker-Shaped Discounted Gain (RSDG), computed from reranker scores over the retriever's top-k documents. For query expansion, RAQE generates a single, short expansion in one pass, avoiding long outputs and repeated generation while remaining low latency. Across eight IR benchmarks, RAQE is competitive with strong large language model prompting baselines and narrows the gap to supervised reinforcement learning methods without relevance labels.

 RAQE: Reranker-Aligned Query Expansion via Label-Free Group-Relative Policy Optimization

Adapting Large Language Models to the medical domain remains an active area of research, with multiple strategies proposed to leverage annotated and unannotated data effectively. In this work, we propose a thesis outline to compare three common adaptation approaches—Instruction Tuning, Continual Pretraining, and Reasoning-oriented Training. We identify $5$ dimensions to analyse: i) the interaction between the adaptation technique and the tasks; ii) the impact of the data size on the downstream performance; iii) the differences between datasets required by the three techniques; iv) the impact of the techniques given the model size; v) the impact of the techniques given the language. We construct an evaluation framework composed by $5$ multilingual medical NLP tasks (named entity recognition, relation extraction, question answering, case report form filling, argument mining), spanning on $21$ datasets in English, Italian, and Spanish, for a total of $61$ combinations of language and sub-task.

 Thesis Proposal: LLMs post-training for multilingual medical tasks. Instruction-Tuning, Continual-Pretraining or Reasoning?

We find that LoRA fine-tuning exhibits un-learning on contested examples: items with high annotator disagreement show increasing loss during training, a qualitatively distinct pattern largely absent under full fine-tuning and consistent across all six models tested (four encoder, two decoder-only). This discovery emerges from correlating annotation entropy, computed from ChaosNLI's 100 labels per example, with per-example area under the loss curve (AULC) on SNLI and MNLI. The correlation is positive in all 25 conditions tested (Spearman $\rho = 0.06$-$0.43$), with decoder-only models showing stronger correlations than encoders at matched LoRA rank. The effect survives partial-correlation controls and replicates across seeds and datasets. A preliminary noise-injection experiment is consistent with these findings.

 Annotation Entropy Predicts Per-Example Learning Dynamics in LoRA Fine-Tuning

Large language models can produce fluent but unfaithful text, known as hallucinations, because point predictions carry no indication of their own reliability. A model that ``knows what it does not know'' could flag, locate, or avoid such failures, yet standard training and decoding offer no such mechanism. Uncertainty is the missing signal, and once calibrated it can serve as a selection filter, an error indicator, or a curriculum guide, three roles unified under a single adaptive-control principle. This thesis develops a framework in which uncertainty is measured from model predictions, calibrated via conformal prediction or analogous principled boundaries, and integrated into the learning or generation process itself. We instantiate this framework in three progressively challenging scenarios: classification under label noise (uncertainty as selection), video generation (uncertainty as error correction), and text generation (uncertainty as curriculum). The principle is validated in classification and video generation and extended to text generation, showing that the pipeline structure is invariant while each component takes qualitatively new forms under escalating difficulty.

Thesis Proposal: Uncertainty as Adaptive Control: From Selection to Curriculum via Conformal Calibration

One partner says "Fine" meaning resolution; the other hears surrender. The word is shared; the affective uptake is not. We formalize this as **affective meaning divergence** (AMD), the total-variation distance between interlocutors' anchor-conditioned affect distributions. Building on speech-act theory, common-ground accumulation, and entropy-regularized game theory, we derive a logit best-response map whose dynamics undergo a *saddle-node bifurcation*: when $\beta\alpha > 4$, a monotone increase in AMD-driven load produces an abrupt, hysteretic collapse of repair coordination. On Conversations Gone Awry (CGA-Wiki; $N=652$), derailing conversations exhibit critical-slowing-down (CSD) signatures across multiple levels: lexical divergence variance ($p<0.001$, $d=0.36$), AMD variance ($p=0.001$, $d=0.26$), and dialog-act repair variance ($p=0.016$, $d=0.20$), all significant after correction and stronger than toxicity and sentiment baselines. AMD provides a distinct temporal signature, with retrospectively measured variance peaking at the bifurcation point while toxicity variance peaks earlier, and is the only indicator grounded in the theoretical framework. Boundary-condition analysis on CGA-CMV ($N=1,169$) yields mixed but directionally consistent evidence.

Phase Transitions in Affective Meaning Divergence: The Hidden Drift Before the Break

Reasoning models frequently agree with incorrect user suggestions - a behavior known as sycophancy. However, it is unclear where in the reasoning trace this agreement originates and how strong the commitment is. We introduce sycophantic anchors - sentences identified via counterfactual analysis that commit models to user agreement. Across four reasoning models spanning three architecture families (Llama, Qwen, Falcon-hybrid) and 1.5B - 8B parameters, we analyze over 200,000 counterfactual rollouts and show that linear probes reliably detect sycophantic anchors (74 - 85\% balanced accuracy), outperforming text-only baselines at high commitment levels - confirming they capture internal states beyond surface vocabulary. Regressors further predict commitment strength from activations ($R^2$ up to 0.74). We observe a consistent asymmetry: sycophancy leaves a stronger mechanistic footprint than correct reasoning. We also find that sycophancy builds gradually during generation rather than being determined by the prompt. These findings enable sentence-level detection and quantification of model misalignment mid-inference.

Sycophantic Anchors: Localizing and Quantifying User Agreement in Reasoning Models

Hate speech detection is essential for maintaining healthy online communities. Large language models (LLMs) perform well on text classification, yet their decision strategies need to be better understood. While post-hoc rationales can justify individual decisions, they substantially increase inference cost and limit scalability in high-throughput settings. As another approach, we propose an extended rational inattention model that parameterizes linguistic noise and information processing cost, providing an interpretable behavioral framework for black-box LLM classifiers. Treating LLMs as rational decision-makers under information constraints allows us to estimate - from the observed classification behavior - the parameters that represent information processing cost and noise sensitivity. As a case study and using a hate-speech dataset spanning multiple noise environments, we evaluate four commercial LLMs and show that the introduced extended rational inattention model predictions closely match the observed performance across different noise levels. We further test the performance under various noise mechanisms and find that the inferred information cost parameters remain consistent while the noise parameters vary with the distortion mechanism. Overall, our introduced framework offers a cost-efficient and quantitative approach to derive interpretable indices of LLM moderation behavior and decisions, without additional rationale generation.

Interpretability of LLM Classifiers via the Rational Inattention Theory with Application to Hate Speech Detection

Adversarial perturbations in the context of large language models (LLMs) are subtle changes added to input data (i.e., images or text) that are designed to alter predictions or outputs of machine learning models. We introduce several novel visualizations using topological data analysis (TDA) (leveraging persistent homology) to characterize how adversarial perturbations act on text inputs, specifically, how sandbagging and code-injection attacks alter the geometric structure of attention heads in transformer models. By computing persistent homology metrics from attention maps across different model architectures (such as BERT, RoBERTa, ELECTRA, DistilGPT, etc.), we find that adversarial inputs alter higher-dimensional topological features ($H_{1}$ loops and $H_{2}$ voids) in ways that distinguish them from clean, non-adversarial inputs.

The Shape of Vulnerability: How Adversarial Perturbations Reshape the Topology of Language Model Latent Spaces

We propose a novel approach to translating Japanese slides into English and to correcting their layout errors by utilizing multimodal LLMs with slide images and XML structures. Existing translation tools often suffer from layout errors after translation due to text expansion during the translation process, causing text to overlap with figures or other items in slides and thereby reducing readability. To overcome this issue, our proposed framework introduces two steps consisting of (i) translating text fragments within the slide, and (ii) correcting layout errors by optimizing layout placement based on visual consistency. In step (ii), we empirically show that few-shot prompts are quite effective in layout error correction. Given that the optimal combination of steps (i) and (ii) varies depending on the slide layout, our method generates eight different layout candidates. Consequently, we introduce a third step that automatically selects the optimal output from these eight candidates. The experimental results showed that the proposed method outperformed baselines and achieved 4.1% layout error rate and over 80% model selection success rate.

Downloads

Next from ACL 2026

Detecting Hallucinations in Large Language Models via Internal Attention Divergence Signals

Similar lecture

The Dominance of Text Space: Unveiling the Asymmetric Nature of Cross-Modal Alignment in Large Language Models

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from ACL 2026

Detecting Hallucinations in Large Language Models via Internal Attention Divergence Signals

Similar lecture

The Dominance of Text Space: Unveiling the Asymmetric Nature of Cross-Modal Alignment in Large Language Models

Downloads