Multi-Agent Path Finding (MAPF) poses a significant and challenging problem critical for applications in robotics and logistics, particularly due to its combinatorial complexity and the partial observability inherent in realistic environments. Decentralized reinforcement learning methods commonly encounter two substantial difficulties: first, they often yield self-centered behaviors among agents, resulting in frequent collisions, and second, their reliance on complex communication modules leads to prolonged training times, sometimes spanning weeks. To address these challenges, we propose an efficient decentralized planning framework based on the Decision Transformer (DT), uniquely leveraging offline reinforcement learning to substantially reduce training durations from weeks to mere hours. Crucially, our approach effectively handles long-horizon credit assignment and significantly improves performance in scenarios with sparse and delayed rewards. Furthermore, to overcome adaptability limitations inherent in standard RL methods under dynamic environmental changes, we integrate a large language model (GPT-4o) to dynamically guide agent policies. Extensive experiments in both static and dynamically changing environments demonstrate that our DT-based approach, augmented briefly by GPT-4o, significantly enhances adaptability and performance.

This paper introduces FoodSEM, a state-of-the-art fine-tuned open-source large language model (LLM) for named-entity linking (NEL) to food-related ontologies. To the best of our knowledge, food NEL is a task that cannot be accurately solved by state-of-the-art general-purpose (large) language models or custom domain-specific models/systems. Through an instruction-response (IR) scenario, FoodSEM links food-related entities mentioned in a text to several ontologies, including FoodOn, SNOMED-CT, and the Hansard taxonomy. The FoodSEM model achieves state-of-the-art performance compared to related models/systems, with F1 scores even reaching 98% on some ontologies and datasets. The presented comparative analyses against zero-shot, one-shot, and few-shot LLM prompting baselines further highlight FoodSEM's superior performance over its non-fine-tuned version. By making FoodSEM and its related resources publicly available, the main contributions of this article include (1) publishing a food-annotated corpora into an IR format suitable for LLM fine-tuning/evaluation, (2) publishing a robust model to advance the semantic understanding of text in the food domain, and (3) providing a strong baseline on food NEL for future benchmarking.

This survey examines multilingual vision-language models that process text and images across languages. We review 31 models and 21 benchmarks, spanning encoder-only and generative architectures, and identify a key tension between language neutrality (consistent cross-lingual representations) and cultural awareness (adaptation to cultural contexts). Current training methods favor neutrality through contrastive learning, while cultural awareness depends on diverse data. Two-thirds of evaluation benchmarks use translation-based approaches prioritizing semantic consistency, though recent work incorporates culturally grounded content. We find discrepancies in cross-lingual capabilities and gaps between training objectives and evaluation goals.

Legal Article Prediction (LAP) is a critical task in legal text classification, leveraging natural language processing (NLP) techniques to automatically predict relevant legal articles based on the fact descriptions of cases. As a foundational step in legal decision-making, LAP plays a pivotal role in determining subsequent judgments, such as charges and penalties. Despite its importance, existing methods face significant challenges in addressing the complexities of LAP. Supervised classification models (SCMs), such as CNN and BERT, struggle to fully capture intricate fact patterns due to their inherent limitations. Conversely, large language models (LLMs), while excelling in generative tasks, perform suboptimally in predictive scenarios due to the abstract and ID-based nature of legal articles. Furthermore, the diversity of legal systems across jurisdictions exacerbates the issue, as most approaches are tailored to specific countries and lack broader applicability. To address these limitations, we propose Uni-LAP, a universal framework for legal article prediction that integrates the strengths of SCMs and LLMs through tight collaboration. Specifically, in Uni-LAP, the SCM is enhanced with a novel Top-K loss function to generate accurate candidate articles, while the LLM employs syllogism-inspired reasoning to refine the final predictions. We evaluated Uni-LAP on datasets from multiple jurisdictions, and empirical results demonstrate that our approach consistently outperforms existing baselines, showcasing its effectiveness and generalizability.

Critic-free methods like GRPO reduce memory demands by estimating advantages from multiple rollouts but tend to converge slowly, as critical learning signals are diluted by an abundance of uninformative samples and tokens. To tackle this challenge, we propose the \textbf{Dynamic Dual-Level Down-Sampling (D$^3$S)} framework that prioritizes the most informative samples and tokens across groups to improve the efficient of policy optimization. D$^3$S operates along two levels: (1) the sample-level, which selects a subset of rollouts to maximize advantage variance ($\text{Var}(A)$). We theoretically proven that this selection is positively correlated with the upper bound of the policy gradient norms, yielding higher policy gradients. (2) the token-level, which prioritizes tokens with a high product of advantage magnitude and policy entropy ($|A_{i,t}|\times H_{i,t}$), focusing updates on tokens where the policy is both uncertain and impactful. Moreover, to prevent overfitting to high-signal data, D$^3$S employs a dynamic down-sampling schedule inspired by curriculum learning. This schedule starts with aggressive down-sampling to accelerate early learning and gradually relaxes to promote robust generalization. Extensive experiments on Qwen2.5 and Llama3.1 demonstrate that integrating D$^3$S into advanced RL algorithms achieves state-of-the-art performance and generalization while requiring \textit{fewer} samples and tokens across diverse reasoning benchmarks. Our code is added in the supplementary materials and will be made publicly available.

Fact-checking real-world claims, particularly numerical claims, is inherently complex that require multistep reasoning and numerical reasoning for verifying diverse aspects of the claim. Although large language models (LLMs) including reasoning models have made tremendous advances, they still fall short on fact-checking real-world claims that require a combination of compositional and numerical reasoning. They are unable to understand nuance of numerical aspects, and are also susceptible to the reasoning drift issue, where the model is unable to contextualize diverse information resulting in misinterpretation and backtracking of reasoning process. In this work, we systematically explore scaling test-time compute (TTS) for LLMs on the task of fact-checking complex numerical claims, which entails eliciting multiple reasoning paths from an LLM. We train a verifier model (VERIFIERFC) to navigate this space of possible reasoning paths and select one that could lead to the correct verdict. We observe that TTS helps mitigate the reasoning drift issue, leading to significant performance gains for fact-checking numerical claims. To improve compute efficiency in TTS, we introduce an adaptive mechanism that performs TTS selectively based on the perceived complexity of the claim. This approach achieves 1.8x higher efficiency than standard TTS, while delivering a notable 18.8% performance improvement over single-shot claim verification methods. Our code and data can be found at https://github.com/VenkteshV/VerifierFC

Large language model (LLM) powered code agents are rapidly transforming software engineering by automating tasks such as testing, debugging, and repairing, yet the security risks of their generated code have become a critical concern. Existing benchmarks have offered valuable insights but remain insufficient: they often overlook the genuine context in which vulnerabilities were introduced or adopt narrow evaluation protocols that fail to capture either functional correctness or newly introduced vulnerabilities. We therefore introduce SecureAgentBench, a benchmark of 105 coding tasks designed to rigorously evaluate code agents' capabilities in secure code generation. Each task includes (i) realistic task settings that require multi-file edits in large repositories, (ii) aligned contexts based on real-world open-source vulnerabilities with precisely identified introduction points, and (iii) comprehensive evaluation that combines functionality testing, vulnerability checking through proof-of-concept exploits, and detection of newly introduced vulnerabilities using static analysis. We evaluate three representative agents (SWE-agent, OpenHands, and Aider) with three state-of-the-art LLMs (Claude 3.7 Sonnet, GPT-4.1, and DeepSeek-V3.1). Results show that (i) current agents struggle to produce secure code, as even the best-performing one, SWE-agent supported by DeepSeek-V3.1, achieves merely 15.2% correct-and-secure solutions, (ii) some agents produce functionally correct code but still introduce vulnerabilities, including new ones not previously recorded, and (iii) adding explicit security instructions for agents does not significantly improve secure coding, underscoring the need for further research. These findings establish SecureAgentBench as a rigorous benchmark for secure code generation and a step toward more reliable software development with LLMs.

Non-English dialogue datasets are scarce, and models are often trained or evaluated on translations of English-language dialogues, an approach which can introduce artifacts that reduce their naturalness and cultural appropriateness. This work proposes Dialogue Act Script (DAS), a structured framework for encoding, localizing, and generating multilingual dialogues from abstract intent representations. Rather than translating dialogue utterances directly, DAS enables the generation of new dialogues in the target language that are culturally and contextually appropriate. By using structured dialogue act representations, DAS supports flexible localization across languages, mitigating translationese and enabling more fluent, naturalistic conversations. Human evaluations across Italian, German, and Chinese show that DAS-generated dialogues consistently outperform those produced by both machine and human translators on measures of cultural relevance, coherence, and situational appropriateness.

Activation steering is a promising technique for controlling LLM behavior by adding semantically meaningful vectors directly into a model's hidden states during inference. It is often framed as a precise, interpretable, and potentially safer alternative to fine-tuning. We demonstrate the opposite: steering systematically breaks model alignment safeguards, making it comply with harmful requests. Through extensive experiments on different model families, we show that even steering in a random direction can increase the probability of harmful compliance from 0% to 2-27%. Alarmingly, steering benign features from a sparse autoencoder (SAE), a common source of interpretable directions, increases these rates by a further 2-4%. Finally, we show that combining 20 randomly sampled vectors that jailbreak a single prompt creates a universal attack, significantly increasing harmful compliance on unseen requests. These results challenge the paradigm of safety through interpretability, showing that precise control over model internals does not guarantee precise control over model behavior.

Prior work has shown that two NLP evaluation experiments that report results for the same quality criterion name (e.g. Fluency) do not necessarily evaluate the same aspect of quality, and the comparability implied by the name can be misleading. Not knowing when two evaluations are comparable in this sense means we currently lack the ability to draw reliable conclusions about system quality on the basis of multiple, independently conducted evaluations. This in turn hampers the ability of the field to progress scientifically as a whole, a pervasive issue in NLP since its beginning (Sparck Jones, 1981). It is hard to see how the issue of unclear comparability can be fully addressed other than by the creation of a standard set of quality criterion names and definitions that the several hundred quality criterion names actually in use in the field can be mapped to, and grounded in. Taking a strictly descriptive approach, the QCET Quality Criteria for Evaluation Taxonomy derives a standard set of quality criterion names and definitions from three surveys of evaluations reported in NLP, and structures them into a hierarchy where each parent node captures common aspects of its child nodes. We present QCET and the resources it consists of, and discuss its three main uses in (i) establishing comparability of existing evaluations, (ii) guiding the design of new evaluations, and (iii) assessing regulatory compliance.