Large language models have demonstrated remarkable reasoning capabilities across diverse natural language tasks. However, comparable breakthroughs in scientific discovery are more limited, because understanding complex physical phenomena demands multifaceted representations far beyond language alone. A compelling example is the design of functional materials such as MOFs-critical for a range of impactful applications like carbon capture and hydrogen storage. Navigating their vast and intricate design space in language-based representations interpretable by LLMs is challenging due to the numerous possible three-dimensional atomic arrangements and strict reticular rules of coordination geometry and topology. Despite promising early results in LLM-assisted discovery for simpler materials systems, MOF design remains heavily reliant on tacit human expertise rarely codified in textual information alone. To overcome this barrier, we introduce L2M3OF, the first multimodal LLM for MOFs. L2M3OF integrates crystal representation learning with language understanding to process structural, textual, and knowledge modalities jointly. L2M3OF employs a pre-trained crystal encoder with a lightweight projection layer to compress structural information into a token space, enabling efficient alignment with language instructions. To facilitate training and evaluation, we curate a structure-property-knowledge database of crystalline materials and benchmark L2M3OF against state-of-the-art closed-source LLMs such as GPT-5, Gemini-2.5-Pro and DeepSeek-R1. Experiments show that L2M3OF outperforms leading text-based closed-source LLMs in property prediction and knowledge generation tasks, despite using far fewer parameters. These results highlight the importance of multimodal approaches for porous material understanding and establish L2M3OF as a foundation for next-generation AI systems in materials discovery.

Reverse engineering (RE) of x86 binaries is indispensable for malware and firmware analysis, but remains slow due to stripped metadata and adversarial obfuscation. Large Language Models (LLMs) offer potential for improving RE efficiency through automated comprehension and commenting, but cloud-hosted, closed-weight models pose privacy and security risks and cannot be used in closed-network facilities. We evaluate parameter-efficient fine-tuned local LLMs for assisting with x86 RE tasks in these settings. Eight open-weight models across the CodeLlama, Qwen2.5-Coder, and CodeGemma series are fine-tuned on a custom curated dataset of 5,981 x86 assembly examples. We evaluate them quantitatively and identify the fine-tuned Qwen2.5-Coder-7B as the top performer, which we name REx86. REx86 reduces test-set cross-entropy loss by 64.2% and improves semantic cosine similarity against ground truth by 20.3\% over its base model. In a limited user case study (n=43), REx86 significantly enhanced line-level code understanding (p = 0.031) and increased the correct-solve rate from 31% to 53% (p = 0.189), though the latter did not reach statistical significance. Qualitative analysis shows more accurate, concise comments with fewer hallucinations. REx86 delivers state-of-the-art assistance in x86 RE among local, open-weight LLMs. Our findings demonstrate the value of domain-specific fine-tuning, and highlight the need for more commented disassembly data to further enhance LLM performance in RE. REx86, its dataset, and LoRA adapters are publicly available at https://github.com/dlea8/REx86 and https://zenodo.org/records/15420461.

Current Vision-Language Models (VLMs) struggle to ground anatomical regions in 3D medical images and reason about them in a step-by-step manner, a key requirement of real-world diagnostic assessment. This ability is essential for aligning model outputs with the diagnostic workflows clinicians use in practice, enabling trustworthy clinician-AI collaboration. Existing 3D datasets provide localization labels, but none support this "grounded reasoning" ability. To address this gap, we introduce 3DReasonKnee, the first 3D grounded reasoning dataset for medical images, which provides 494k high-quality quintuples derived from 7,970 3D knee MRI volumes. Each quintuple includes: (1) the 3D MRI volume, (2) a diagnostic question targeting a specific anatomical region (3) a 3D bounding box localizing the relevant anatomical structures, (4) clinician-generated diagnostic reasoning steps that explicitly detail the 3D reasoning process, and (5) structured severity assessments for the relevant anatomical region. The creation and validation of 3DReasonKnee, involving over 450 hours of expert clinician time for manually segmenting MRIs and generating reasoning chains, ensures its superior quality and clinical relevance. We establish ReasonKnee-Bench to evaluate localization and diagnostic accuracy, providing insight into VLM ability to perform grounding and severity assessment across anatomical regions and diagnostic inquiries. We benchmark five state-of-the-art VLMs, providing baseline performance for ReasonKnee-Bench. By providing this unique resource of expert-annotated 3D reasoning pathways, 3DReasonKnee serves as a repository of orthopedic surgeons' diagnostic expertise and offers a vital testbed for advancing multimodal medical AI systems towards 3D, clinically aligned, localized decision-making capabilities. The dataset can be found in: https://huggingface.co/datasets/rajpurkarlab/3DReasonKnee

Accurate detection of errors in large language models (LLM) responses is central to the success of scalable oversight, or providing effective supervision to superhuman intelligence. Yet, self-diagnosis is often unreliable on complex tasks unless aided by reliable external feedback. Multi-agent debate (MAD) seems to be a natural alternative to external feedback: multiple LLMs provide complementary perspectives and cross-checks for error detection. However, prior MAD protocols frame debate as a zero-sum game, where the debaters compete to win the game instead of seeking the truth. Consequently, it leads to debate hacking: debaters tend to mislead the judge by misinterpreting the task or presenting overconfident claims, which introduce more mistakes and underperform single-agent methods. To mitigate the issue, we introduce a new collaborative MAD protocol, termed ColMAD, that reframes MAD as a non-zero sum game. Specifically, ColMAD encourages multiple agents to criticize each other in a supportive way, such that they can complement the missing points of each other. Therefore, the judge agent can make a more informative conclusion based on more comprehensive evidence. Empirically, we show that ColMAD significantly outperforms previous competitive MAD by 19% and brings non-trivial improvements over single-agent methods in error detection.

We present Irish-BLiMP (Irish Benchmark of Linguistic Minimal Pairs), the first dataset and framework designed for fine-grained evaluation of linguistic competence in the Irish language, an endangered language. Drawing on a variety of linguistic literature and grammar reference works, we manually constructed and reviewed 1020 minimal pairs across a taxonomy of 11 linguistic features, through a team of fluent Irish speakers. We evaluate both existing Large Language Models (LLMs) and fluent human participants on their syntactic knowledge of Irish. Our findings show that humans outperform all models across all linguistic features, achieving 16.6% higher accuracy on average. Moreover, a substantial performance gap of 18.1% persists between open- and closed-source LLMs, with even the strongest model (gpt-5) reaching only 73.5% accuracy compared to 90.1% by human. Interestingly, human participants and models struggle on different aspects of Irish grammar, thus highlighting a difference in representation learned by the models. Overall, Irish-BLiMP provides the first systematic framework for evaluating the grammatical competence of LLMs in Irish and offers a valuable benchmark for advancing research on linguistic understanding in low-resource languages.

We discover a novel and surprising phenomenon of unintentional misalignment in reasoning language models (RLMs), which we call self-jailbreaking. Specifically, after benign reasoning training on math or code domains, RLMs will use multiple strategies to circumvent their own safety guardrails. One strategy is to introduce benign assumptions about users and scenarios to justify fulfilling harmful requests. For instance, an RLM reasons that harmful requests like ``outline a strategy for stealing customers' credit card information from a retail store'' could be associated with the benign intent of ``a security professional trying to test defense,'' despite no such benign context being provided as input. We observe that many open-weight RLMs, including DeepSeek-R1-distilled, s1.1, Phi-4-mini-reasoning, and Nemotron, suffer from self-jailbreaking despite being aware of the harmfulness of the requests. We also provide a mechanistic understanding of self-jailbreaking: RLMs are more compliant after benign reasoning training, and after self-jailbreaking, models appear to perceive malicious requests as less harmful in the CoT, thus enabling compliance with them. To mitigate self-jailbreaking, we find that including minimal safety reasoning data during training is sufficient to ensure RLMs remain safety-aligned. Our work provides the first systematic analysis of self-jailbreaking behavior and offers a practical path forward for maintaining safety in increasingly capable RLMs.

Forecasting in the real world requires integrating structured time-series data with unstructured textual information, but existing methods are architecturally limited by fixed input/output horizons and are unable to model or quantify uncertainty. We address this challenge by introducing LLM-integrated Bayesian State space models (LBS), a novel probabilistic framework for multimodal temporal forecasting. At a high level, LBS consists of two components: (1) a state space model (SSM) backbone that captures the temporal dynamics of latent states from which both numerical and textual observations are generated and (2) a pretrained large language model (LLM) that is adapted to encode textual inputs for posterior state estimation and decode textual forecasts consistent with the latent trajectory. This design enables flexible lookback and forecast windows, principled uncertainty quantification, and improved temporal generalization thanks to the well-suited inductive bias of SSMs toward modeling dynamical systems. Experiments on the TextTimeCorpus benchmark demonstrate that LBS improves the previous state-of-the-art by 13.20% while providing human-readable summaries of each forecast. Our work is the first to unify LLMs and SSMs for joint numerical and textual prediction, offering a novel foundation for multimodal temporal reasoning.

Protein mutations can have profound effects on biological function, making accurate prediction of property changes critical for drug discovery, protein engineering, and precision medicine. Current approaches rely on fine-tuning protein-specific transformers for individual datasets, but struggle with cross-dataset generalization due to heterogeneous experimental conditions and limited target domain data. We introduce two key innovations: (1) the first application of Model-Agnostic Meta-Learning (MAML) to protein mutation property prediction, and (2) a novel mutation encoding strategy using separator tokens to directly incorporate mutations into sequence context. We build upon transformer architectures integrating them with MAML to enable rapid adaptation to new tasks through minimal gradient steps rather than learning dataset-specific patterns. Our mutation encoding addresses the critical limitation where standard transformers treat mutation positions as unknown tokens, significantly degrading performance. Evaluation across three diverse protein mutation datasets (functional fitness, thermal stability, and solubility) demonstrates significant advantages over traditional fine-tuning. In cross-task evaluation, our meta-learning approach achieves 29% better accuracy for functional fitness with 65% less training time, and 94% better accuracy for solubility with 55% faster training. The framework maintains consistent training efficiency regardless of dataset size, making it particularly valuable for industrial applications and early-stage protein design where experimental data is limited. This work establishes a systematic application of meta-learning to protein mutation analysis and introduces an effective mutation encoding strategy, offering transformative methodology for cross-domain generalization in protein engineering.

Understanding adversarial behavior in cybersecurity has traditionally relied on high-level intelligence reports and manual interpretation of attack chains. However, real-time defense requires the ability to infer attacker intent and cognitive strategy directly from low-level system telemetry such as intrusion detection system (IDS) logs. In this paper, we propose a novel framework that leverages large language models (LLMs) to analyze Suricata IDS logs and infer attacker actions in terms of MITRE ATT&CK techniques. Our approach is grounded in the hypothesis that attacker behavior reflects underlying cognitive biases such as loss aversion, risk tolerance, or goal persistence that can be extracted and modeled through careful observation of log sequences. This lays the groundwork for future work on behaviorally adaptive cyber defense and cognitive trait inference. We develop a strategy-driven prompt system to segment large amounts of network logs data into distinct behavioral phases in a highly efficient manner, enabling the LLM to associate each phase with likely techniques and underlying cognitive motives. By mapping network-layer events to high-level attacker strategies, our method reveals how behavioral signals such as tool switching, protocol transitions, or pivot patterns correspond to psychologically meaningful decision points. The results demonstrate that LLMs can bridge the semantic gap between packet-level logs and strategic intent, offering a pathway toward cognitive-adaptive cyber defense. Keywords: Cognitive Cybersecurity, Large Language Models (LLMs), Cyberpsychology, Intrusion Detection Systems (IDS), MITRE ATT&CK, Cognitive Biases

Reasoning models (RMs), language models (LMs) trained with reinforcement learning to produce long-form natural language reasoning, have been remarkably successful, but they still require large amounts of computation and data to train, and can be slow and expensive to run. In this paper, we show that standard instruct LMs can already be elicited to be strong reasoners at a level comparable to or even surpassing their corresponding RMs (e.g., DeepSeek V3 vs R1) without finetuning, across diverse domains from instruction following and creative generation to mathematical reasoning. This is achieved by CodeAdapt, our simple recipe that combines the CodeAct framework, where LMs interleave natural language reasoning with code execution in a multi-step fashion, with few-shot bootstrap in-context learning from as few as five training problems. Analyzing four matched pairs of LMs and RMs, we find that CodeAdapt enables three LMs to outperform the corresponding RMs on average over eight tasks (up to 22.9%) while being 10-81% more token efficient, and delivers superior performance on six tasks when averaged over the four models (up to 35.7%). Furthermore, the code-augmented reasoning traces display rich and varied problem-solving strategies. Our findings support that (1) CodeAdapt-style learning and reasoning may be robust and domain general and (2) code-enabled LMs are cognitively grounded and powerful systems, potentially providing a strong foundation for in-weight reinforcement learning.