Just as humans display language patterns influenced by their native tongue when speaking new languages, LLMs often default to English-centric responses even when generating in other languages. Nevertheless, we observe that local cultural information persists within the models and can be readily activated for cultural customization. We first demonstrate that explicitly providing cultural context in prompts significantly improves the models' ability to generate culturally localized responses. We term the disparity in model performance with versus without explicit cultural context the explicit-implicit localization gap, indicating that while cultural knowledge exists within LLMs, it may not naturally surface in multilingual interactions if cultural context is not explicitly provided. Despite the explicit prompting benefit, however, the answers reduce in diversity and tend toward stereotypes. Second, we identify an explicit cultural customization vector, conserved across all non-English languages we explore, which enables LLMs to be steered from the synthetic English cultural world-model toward each non-English cultural world. Steered responses retain the diversity of implicit prompting and reduce stereotypes to dramatically improve the potential for customization. We discuss the implications of explicit cultural customization for understanding the conservation of alternative cultural world models within LLMs, and their controllable utility for translation, cultural customization, and the possibility of making the explicit implicit through soft control for expanded LLM function and appeal.

Multimodal Large Language Models (MLLMs) are set to transform how machines process and generate human-like responses by integrating diverse modalities such as text, images, and code. Yet, effectively harnessing their capabilities hinges on optimal prompt engineering. We present a comprehensive experimental evaluation of seven prompt engineering methods applied to 13 open-source MLLMs over 24 tasks spanning Reasoning and Compositionality, Multimodal Understanding and Alignment, Complex Code Generation and Execution, and Knowledge Retrieval and Integration. Our approach stratifies models by parameter count into Small (<4B), Medium (4B-10B), and Large (>10B) categories and compares prompting techniques including Zero-Shot, One-Shot, Few-Shot, Chain-of-Thought, Analogical, Generated Knowledge, and Tree-of-Thought. While Large MLLMs excel in structured tasks such as code generation, achieving accuracies up to 96.88% under Few-Shot prompting, all models struggle with complex reasoning and abstract understanding, often yielding accuracies below 60% and high hallucination rates. Structured reasoning prompts frequently increased hallucination up to 75% in small models and led to longer response times (over 20 seconds in Large MLLMs), while simpler prompting methods provided more concise and efficient outputs. No single prompting method uniformly optimises all task types. Instead, adaptive strategies combining example-based guidance with selective structured reasoning are essential to enhance robustness, efficiency, and factual accuracy. Our findings offer practical recommendations for prompt engineering and support more reliable deployment of MLLMs across applications including AI-assisted coding, knowledge retrieval, and multimodal content understanding.

In this paper, we present HalluSearch, a multilingual pipeline designed to detect fabricated text spans in Large Language Model (LLM) outputs. Developed as part of Mu-SHROOM, the Multilingual Shared-task on Hallucinations and Related Observable Overgeneration Mistakes, HalluSearch couples retrieval-augmented verification with fine-grained factual splitting to identify and localize hallucinations in fourteen different languages. Empirical evaluations show that HalluSearch performs competitively, placing fourth in both English (within the top ten percent) and Czech. While the system's retrieval-based strategy generally proves robust, it faces challenges in languages with limited online coverage, underscoring the need for further research to ensure consistent hallucination detection across diverse linguistic contexts.

Despite the rapid advancement of large language models, they remain highly susceptible to generating hallucinations, which significantly hinders their widespread application. Hallucination research requires dynamic and fine-grained evaluation. However, most existing hallucination benchmarks (especially in Chinese language) rely on human annotations, making automatical and cost-effective hallucination evaluation challenging. To address this, we introduce HaluAgent, an agentic framework that automatically constructs fine-grained QA dataset based on some knowledge documents. Our experiments demonstrate that the manually designed rules and prompt optimization can improve the quality of generated data. Using HaluAgent, we construct C-FAITH, a Chinese QA hallucination benchmark created from 1,399 knowledge documents obtained from web scraping, totaling 60,702 entries. We comprehensively evaluate 16 mainstream LLMs with our proposed C-FAITH, providing detailed experimental results and analysis.

Large-scale reinforcement learning (RL) methods have proven highly effective in enhancing the reasoning abilities of large language models (LLMs), particularly for tasks with verifiable solutions such as mathematics and coding. However, applying this idea to machine translation (MT), where outputs are flexibly formatted and difficult to automatically evaluate with explicit rules, remains underexplored. In this work, we introduce MT-R1-Zero, the first open-source adaptation of the R1-Zero RL framework for MT without supervised fine-tuning or cold-start. We propose a rule-metric mixed reward mechanism to guide LLMs towards improved translation quality via emergent reasoning. On the WMT 24 English-Chinese benchmark, our MT-R1-Zero-3B-Mix achieves competitive performance, surpassing TowerInstruct-7B-v0.2 by an average of 1.26 points. Meanwhile, our MT-R1-Zero-7B-Mix attains a high average score of 62.25 across all metrics, placing it on par with advanced proprietary models such as GPT-4o and Claude-3.5-Sonnet, while the MT-R1-Zero-7B-Sem variant achieves state-of-the-art scores on semantic metrics. Moreover, our work exhibits strong generalization capabilities on out-of-distribution MT tasks, robustly supporting multilingual and low-resource settings. Extensive analysis of model behavior across different initializations and reward metrics offers pioneering insight into the critical role of reward design, LLM adaptability, training dynamics, and emergent reasoning patterns within the R1-Zero paradigm for MT. Our code is available at https://github.com/fzp0424/MT-R1-Zero.

Current object detectors often suffer significant perfor-mance degradation in real-world applications when encountering distributional shifts. Consequently, the out-of-distribution (OOD) generalization capability of object detectors has garnered increasing attention from researchers. Despite this growing interest, there remains a lack of a large-scale, comprehensive dataset and evaluation benchmark with fine-grained annotations tailored to assess the OOD generalization on more intricate tasks like object detection and grounding. To address this gap, we introduce COUNTS, a large-scale OOD dataset with object-level annotations. COUNTS encompasses 14 natural distributional shifts, over 222K samples, and more than 1,196K labeled bounding boxes. Leveraging COUNTS, we introduce two novel benchmarks: O(OD)2 and OODG. O(OD)2 is designed to comprehensively evaluate the OOD generalization capabilities of object detectors by utilizing controlled distribution shifts between training and testing data. OODG, on the other hand, aims to assess the OOD generalization of grounding abilities in multimodal large language models (MLLMs). Our findings reveal that, while large models and extensive pre-training data substantially en hance performance in in-distribution (IID) scenarios, significant limitations and opportunities for improvement persist in OOD contexts for both object detectors and MLLMs. In visual grounding tasks, even the advanced GPT-4o and Gemini-1.5 only achieve 56.7% and 28.0% accuracy, respectively. We hope COUNTS facilitates advancements in the development and assessment of robust object detectors and MLLMs capable of maintaining high performance under distributional shifts.

Large Reasoning Models (LRMs), such as OpenAI o1 and DeepSeek-R1, have been rapidly progressing and achieving breakthrough performance on complex reasoning tasks such as mathematics and coding. However, the open-source R1 models have raised safety concerns in wide applications, such as the tendency to comply with malicious queries, which greatly impacts the utility of these powerful models in their applications. In this paper, we introduce RealSafe-R1 as safety-aligned versions of DeepSeek-R1 distilled models. To train these models, we construct a dataset of 15k safety-aware reasoning trajectories generated by DeepSeek-R1, under explicit instructions for expected refusal behavior. Both quantitative experiments and qualitative case studies demonstrate the models' improvements, which are shown in their safety guardrails against both harmful queries and jailbreak attacks. Importantly, unlike prior safety alignment efforts that often compromise reasoning performance, our method preserves the models' reasoning capabilities by maintaining the training data within the original distribution of generation. Model weights of RealSafe-R1 are open-source at https://huggingface.co/RealSafe.

Commonsense reasoning deals with the implicit knowledge that is well understood by humans and typically acquired via interactions with the world. In recent times, commonsense reasoning and understanding of various LLMs have been evaluated using text-based tasks. In this work, we argue that a proxy of this understanding can be maintained as a graphical structure that can further help to perform a rigorous evaluation of commonsense reasoning abilities about various real-world activities. We create an annotation scheme for capturing this implicit knowledge in the form of a graphical structure for 37 daily human activities. We find that the created resource can be used to frame an enormous number of commonsense queries (~ 10^{17}), facilitating rigorous evaluation of commonsense reasoning in LLMs. Moreover, recently, the remarkable performance of LLMs has raised questions about whether these models are truly capable of reasoning in the wild and, in general, how reasoning occurs inside these models. In this resource paper, we bridge this gap by proposing design mechanisms that facilitate research in a similar direction. Our findings suggest that the reasoning components are localized in LLMs that play a prominent role in decision-making when prompted with a commonsense query.

An agent's intention often remains hidden behind the black-box nature of embodied policies. Communication using natural language statements that describe the next action can provide transparency towards the agent's behavior. We aim to insert transparent behavior directly into the learning process, by transforming the problem of policy learning into a language generation problem and combining it with traditional autoregressive modelling. The resulting model produces transparent natural language statements followed by tokens representing the specific actions to solve long-horizon tasks in the Language-Table environment. Following previous work, the model is able to learn to produce a policy represented by special discretized tokens in an autoregressive manner. We place special emphasis on investigating the relationship between predicting actions and producing high-quality language for a transparent agent. We find that in many cases both the quality of the action trajectory and the transparent statement increase when they are generated simultaneously.

Vision-Language Models (VLMs) can process visual and textual information in multiple formats: texts, images, interleaved texts and images, or even hour-long videos. In this work, we conduct fine-grained quantitative and qualitative analyses of automatic summarization of multimodal presentations using VLMs with various representations as input. From these experiments, we suggest cost-effective strategies for generating summaries from text-heavy multimodal documents under different input-length budgets using VLMs. We show that slides extracted from the video stream can be beneficially used as input against the raw video, and that a structured representation from interleaved slides and transcript provides the best performance. Finally, we reflect and comment on the nature of cross-modal interactions in multimodal presentations and share suggestions to improve the capabilities of VLMs to understand documents of this nature.