Knowledge of slang is a desirable feature of LLMs in the context of user interaction, as slang often reflects an individual's social identity. Several works on informal language processing have defined and curated benchmarks for tasks such as detection and identification of slang. In this paper, we focus on queer slang. Queer slang can be mistakenly flagged as hate speech or can evoke negative responses from LLMs during user interaction. Research efforts so far have not focused explicitly on queer slang. In particular, detection and processing of queer slang have not been thoroughly evaluated due to the lack of a high-quality annotated benchmark. To address this gap, we curate SLAyiNG, the first dataset containing annotated queer slang derived from subtitles, social media posts, and podcasts, reflecting real-world usage. We describe our data curation process, including the collection of slang terms and definitions, scraping sources for examples that reflect usage of these terms, and our ongoing annotation process. As preliminary results, we calculate inter-annotator agreement for human annotators and OpenAI's model o3-mini, evaluating performance on the task of sense disambiguation. Reaching an average Krippendorff's alpha of 0.746, we argue that state-of-the-art reasoning models can serve as tools for pre-filtering, but the complex and often sensitive nature of queer language data requires expert and community-driven annotation efforts.

Most languages lack sufficient data for large-scale monolingual pretraining, creating a "data wall." Multilingual pretraining helps but is limited by language imbalance and the "curse of multilinguality." An alternative is to translate high-resource text with machine translation (MT), which raises three questions: (1) How does MT-derived data scale with model capacity? (2) Can source-side transformations (e.g., simplifying English with an LLM) improve generalization to native text? (3) How well do models pretrained on MT-derived data adapt when continually trained on limited native text? We investigate these questions by translating English into Indonesian and Tamil--two typologically distant, lower-resource languages--and pretraining GPT-2 models (124M-774M) on native or MT-derived corpora from raw and LLM-simplified English. We evaluate cross-entropy loss on native text, along with accuracy on syntactic probes and downstream tasks. Our results show that (1) MT-pretrained models benefit from scaling; (2) source-side simplification harms generalization to native text; and (3) adapting MT-pretrained models on native text often yields better performance than native-only models, even with less native data. However, tasks requiring cultural nuance (e.g., toxicity detection) demand more exposure to native data.

Training large language models (LLMs) from scratch can yield models with unique functionalities and strengths, but it is costly and often leads to redundant capabilities. A more cost-effective alternative is to fuse existing pre-trained LLMs with different architectures into a more powerful model. However, a key challenge in existing model fusion is their dependence on manually predefined vocabulary alignment, which may not generalize well across diverse contexts, leading to performance degradation in several evaluation. To solve this, we draw inspiration from distribution learning and propose the probabilistic token alignment method as a general and soft mapping for alignment, named as PTA-LLM. Our approach innovatively reformulates token alignment into a classic mathematical problem: optimal transport, seamlessly leveraging distribution-aware learning to facilitate more coherent model fusion. Apart from its inherent generality, PTA-LLM exhibits interpretability from a distributional perspective, offering insights into the essence of the token alignment. Empirical results demonstrate that probabilistic token alignment enhances the target model's performance across multiple capabilities. Our code is avaliable at https://runjia.tech/neurips_pta-llm/.

Low-cost mobile sensors can be used to collect PM$_{2.5}$ concentration data throughout an entire city. However, identifying air pollution hotspots from the data is challenging due to the uneven spatial sampling, temporal variations in the background air quality, and the dynamism of urban air pollution sources. This study proposes a method to identify urban PM$_{2.5}$ hotspots that addresses these challenges, involving four steps: (1) equip citizen scientists with mobile PM$_{2.5}$ sensors while they travel; (2) normalise the raw data to remove the influence of background ambient pollution levels; (3) fit a Gaussian process regression model to the normalised data and (4) calculate a grid of spatially explicit 'hotspot scores' using the probabilistic framework of Gaussian processes, which conveniently summarise the relative pollution levels throughout the city. We apply our method to create the first ever map of PM$_{2.5}$ pollution in Kigali, Rwanda, at a 200m resolution. Our results suggest that the level of ambient PM$_{2.5}$ pollution in Kigali is dangerously high, and we identify the hotspots in Kigali where pollution consistently exceeds the city-wide average. We also evaluate our method using simulated mobile sensing data for Beijing, China, where we find that the hotspot scores are probabilistically well calibrated and accurately reflect the 'ground truth' spatial profile of PM$_{2.5}$ pollution. Thanks to the use of open-source software, our method can be re-applied in cities throughout the world with a handful of low-cost sensors. The method can help fill the gap in urban air quality information and empower public health officials.

A variety of recent methods guide large language model outputs via the inference-time addition of steering vectors to residual-stream or attention-head representations. In contrast, we propose to inject steering vectors directly into the query and value representation spaces within attention heads. We provide evidence that a greater portion of these spaces exhibit high linear discriminability of concepts --a key property motivating the use of steering vectors-- than attention head outputs. We analytically characterize the effect of our method, which we term DISentangled COmmunication (DISCO) Steering, on attention head outputs. Our analysis reveals that DISCO disentangles a strong but underutilized baseline, steering attention inputs, which implicitly modifies queries and values in a rigid manner. In contrast, DISCO's direct modulation of these components enables more granular control. We find that DISCO achieves superior performance over a number of steering vector baselines across multiple datasets on LLaMA 3.1 8B and Gemma 2 9B, with steering efficacy scoring up to 19.1% higher than the runner-up. Our results support the conclusion that the query and value spaces are powerful building blocks for steering vector methods.

Quantifying identity fusion -- the psychological merging of self with another entity or abstract target (e.g., a religious group, political party, ideology, value, brand, belief, etc.) -- is vital for understanding a wide range of group-based human behaviors. We introduce the Cognitive Linguistic Identity Fusion Score (CLIFS), a novel metric that integrates cognitive linguistics with large language models (LLMs), which builds on implicit metaphor detection. Unlike traditional pictorial and verbal scales, which require controlled surveys or direct field contact, CLIFS delivers fully automated, scalable assessments while maintaining strong alignment with the established verbal measure. In benchmarks, CLIFS outperforms both existing automated approaches and human annotation. As a proof of concept, we apply CLIFS to violence risk assessment to demonstrate that it can improve violence risk assessment by more than 240%. Building on our identification of a new NLP task and early success, we underscore the need to develop larger, more diverse datasets that encompass additional fusion-target domains and cultural backgrounds to enhance generalizability and further advance this emerging area. CLIFS models and code are public at https://github.com/DevinW-sudo/CLIFS.

This research examines how Artificial Intelligence (AI) can improve participatory budgeting processes within smart cities. In response to challenges like declining civic participation and resource allocation conflicts, the study explores how online political participation can be improved by AI. It investigates the state capacity governments need to implement AI-enhanced participatory tools, considering technological dependencies and vulnerabilities. It analyzes technological and administrative structures, actors, interests, and strategies to understand the dynamics of online political participation technologies in the case of Sao Paulo, Brazil. The study contributes to understanding how technological advancements can reshape participatory budgeting processes. In a broader sense, the research highlights how AI can transform participatory institutions by offering new tools for citizens and also for government officials in charge of participatory processes within smart cities.

State-of-the-art automatic speech recognition (ASR) models like Whisper, perform poorly on atypical speech, such as that produced by individuals with dysarthria. Past works for atypical speech have mostly investigated fully personalized (or idiosyncratic) models, but modeling strategies that can both generalize and handle idiosyncracy could be more effective for capturing atypical speech. To investigate this, we compare four strategies: (a) $\textit{normative}$ models trained on typical speech (no personalization), (b) $\textit{idiosyncratic}$ models completely personalized to individuals, (c) $\textit{dysarthric-normative}$ models trained on other dysarthric speakers, and (d) $\textit{dysarthric-idiosyncratic}$ models which combine strategies by first modeling normative patterns before adapting to individual speech. In this case study, we find the dysarthric-idiosyncratic model performs better than idiosyncratic approach while requiring less than half as much personalized data (36.43 WER with 128 train size vs 36.99 with 256). Further, we found that tuning the speech encoder alone (as opposed to the LM decoder) yielded the best results reducing word error rate from 71% to 32% on average. Our findings highlight the value of leveraging both normative (cross-speaker) and idiosyncratic (speaker-specific) patterns to improve ASR for underrepresented speech populations.

Cybersecurity threats continue to increase, with a growing number of previously unknown attacks each year targeting both large corporations and smaller entities. This scenario demands the implementation of advanced security measures, not only to mitigate damage but also to anticipate emerging attack trends. In this context, deception tools have become a key strategy, enabling the detection, deterrence, and deception of potential attackers while facilitating the collection of information about their tactics and methods. Among these tools, honeypots have proven their value, although they have traditionally been limited by rigidity and configuration complexity, hindering their adaptability to dynamic scenarios. The rise of artificial intelligence, and particularly general-purpose Large Language Models (LLMs), is driving the development of new deception solutions capable of offering greater adaptability and ease of use. This work proposes the design and implementation of an LLM-based honeypot to simulate an LDAP server, a critical protocol present in most organizations due to its central role in identity and access management. The proposed solution aims to provide a flexible and realistic tool capable of convincingly interacting with attackers, thereby contributing to early detection and threat analysis while enhancing the defensive capabilities of infrastructures against intrusions targeting this service.

As urbanization and climate change progress, urban heat island effects are becoming more frequent and severe. To formulate effective mitigation plans, cities require detailed air temperature data. However, predictive analytics methods based on conventional machine learning models and limited data infrastructure often provide inaccurate predictions, especially in underserved areas. In this context, geospatial foundation models trained on unstructured global data demonstrate strong generalization and require minimal fine-tuning, offering an alternative for predictions where traditional approaches are limited. This study fine-tunes a geospatial foundation model to predict urban land surface temperatures under future climate scenarios and explores its response to land cover changes using simulated vegetation strategies. The fine-tuned model achieved pixel-wise downscaling errors below 1.74 °C and aligned with ground truth patterns, demonstrating an extrapolation capacity up to 3.62 °C.