Security misconfigurations in Container Orchestrators (COs) can pose serious threats to software systems. While Static Analysis Tools (SATs) can effectively detect these security vulnerabilities, the industry currently lacks automated solutions capable of fixing these misconfigurations. The emergence of Large Language Models (LLMs), with their proven capabilities in code understanding and generation, presents an opportunity to address this limitation. This study introduces LLMSecConfig, an innovative framework that bridges this gap by combining SATs with LLMs. Our approach leverages advanced prompting techniques and Retrieval-Augmented Generation (RAG) to automatically repair security misconfigurations while preserving operational functionality. Evaluation of 1,000 real-world Kubernetes configurations achieved a 94\% success rate while maintaining a low rate of introducing new misconfigurations. Our work makes a promising step towards automated container security management, reducing the manual effort required for configuration maintenance.

The enhance of accuracy in reasoning results of LLMs arouses the community's interests, wherein pioneering studies investigate post-hoc strategies to rectify potential mistakes. Despite extensive efforts, they are all stuck in a state of resource competition demanding significant time and computing expenses. The cause of the situation lies in the failure of identifying the fundamental feature of the solutions in this line, coined as the self-denial of LLMs. In other words, LLMs should confidently determine the potential existence of mistakes and carefully execute the targeted correction. As the whole procedure conducts within LLMs, supporting and persuasive references are hard to acquire, while the absence of specific steps towards refining hidden mistakes persists even when errors are acknowledged. In response to the challenges, we present PSSD, which refers to and implements the human psyche structure such that three distinct and interconnected roles contribute to human reasoning. Specifically, PSSD leverages the recent multi-agent paradigm, and is further enhanced with three innovatively conceived roles: (1) the intuition-based id role that provides initial attempts based on benign LLMs; (2) the rule-driven superego role that summarizes rules to regulate the above attempts, and returns specific key points as guidance; and (3) the script-centric ego role that absorbs all procedural information to generate executable script for the final answer prediction. Extensive experiments demonstrate that the proposed design not only better enhance reasoning capabilities, but also seamlessly integrate with current models, leading to superior performance.

Cryptocurrencies are widely used, yet current methods for analyzing transactions heavily rely on opaque, black-box models. These lack interpretability and adaptability, failing to effectively capture behavioral patterns. Many researchers, including us, believe that Large Language Models (LLMs) could bridge this gap due to their robust reasoning abilities for complex tasks. In this paper, we test this hypothesis by applying LLMs to real-world cryptocurrency transaction graphs, specifically within the Bitcoin network. We introduce a three-tiered framework to assess LLM capabilities: foundational metrics, characteristic overview, and contextual interpretation. This includes a new, human-readable graph representation format, LLM4TG, and a connectivity-enhanced sampling algorithm, CETraS, which simplifies larger transaction graphs. Experimental results show that LLMs excel at foundational metrics and offer detailed characteristic overviews. Their effectiveness in contextual interpretation suggests they can provide useful explanations of transaction behaviors, even with limited labeled data.

The rapid advancements in large language models (LLMs), particularly in their reasoning capabilities, hold transformative potential for addressing complex challenges in atmospheric science. However, leveraging LLMs effectively in this domain requires a robust and comprehensive evaluation benchmark. To address this need, we present AtmosSci-Bench, a novel benchmark designed to systematically assess LLM performance across five core categories of atmospheric science problems: hydrology, atmospheric dynamics, atmospheric physics, geophysics, and physical oceanography. We employ a template-based question generation framework, enabling scalable and diverse multiple-choice questions curated from graduate-level atmospheric science problems. We conduct a comprehensive evaluation of representative LLMs, categorized into four groups: instruction-tuned models, advanced reasoning models, math-augmented models, and domain-specific climate models. Our analysis provides some interesting insights into the reasoning and problem-solving capabilities of LLMs in atmospheric science. We believe AtmosSci-Bench can serve as a critical step toward advancing LLM applications in climate service by offering a standard and rigorous evaluation framework. Our source codes are currently available at https://github.com/Relaxed-System-Lab/AtmosSci-Bench.

As large language models (LLMs) are becoming more capable and widespread, the study of their failure cases is becoming increasingly important. Recent advances in standardizing, measuring, and scaling test-time compute suggest new methodologies for optimizing models to achieve high performance on hard tasks. In this paper, we apply these advances to the task of model jailbreaking: eliciting harmful responses from aligned LLMs. We develop an adversarial reasoning approach to automatic jailbreaking via test-time computation that achieves SOTA attack success rates (ASR) against many aligned LLMs, even the ones that aim to trade inference-time compute for adversarial robustness. Our approach introduces a new paradigm in understanding LLM vulnerabilities, laying the foundation for the development of more robust and trustworthy AI systems.

Machine learning models are routinely trained on a mixture of different data domains. Different domain weights yield very different downstream performances. We propose the Soup-of-Experts, a novel architecture that can instantiate a model at test time for any domain weights with minimal computational cost and without re-training the model. Our architecture consists of a bank of expert parameters, which are linearly combined to instantiate one model. We learn the linear combination coefficients as a function of the input domain weights. To train this architecture, we sample random domain weights, instantiate the corresponding model, and backprop through one batch of data sampled with these domain weights. We demonstrate how our approach obtains small specialized models on several language modeling tasks quickly. Soup-of-Experts are particularly appealing when one needs to ship many different specialist models quickly under a model size constraint.

Many commercial Large Language Models (LLMs) are often closed-source, limiting developers to prompt tuning for aligning content generation with specific applications. While these models currently do not provide access to token logits, we argue that if such access were available, it would enable more powerful adaptation techniques beyond prompt engineering. In this paper, we propose a token-level probability reweighting framework that, given access to logits and a small amount of task-specific data, can effectively steer black-box LLMs toward application-specific content generation. Our approach views next-token prediction through the lens of supervised classification. We show that aligning black-box LLMs with task-specific data can be formulated as a label noise correction problem, leading to \emph{Plugin} model -- an autoregressive probability reweighting model that operates solely on logits. We provide theoretical justification for why reweighting logits alone is sufficient for task adaptation. Extensive experiments with multiple datasets, LLMs, and reweighting models demonstrate the effectiveness of our method, advocating for broader access to token logits in closed-source models.

Interpreting and controlling the internal mechanisms of large language models (LLMs) is crucial for improving their trustworthiness and utility. Recent efforts have primarily focused on identifying and manipulating neurons by establishing discrete mappings between neurons and semantic concepts. However, such mappings struggle to handle the inherent polysemanticity in LLMs, where individual neurons encode multiple, distinct concepts. This makes precise control challenging and complicates downstream interventions. Through an in-depth analysis of both encoder and decoder-based LLMs across multiple text classification datasets, we uncover that while individual neurons encode multiple concepts, their activation magnitudes vary across concepts in distinct, Gaussian-like patterns. Building on this insight, we introduce NeuronLens, a novel range-based interpretation and manipulation framework that provides a finer view of neuron activation distributions to localize concept attribution within a neuron. Extensive empirical evaluations demonstrate that NeuronLens significantly reduces unintended interference, while maintaining precise control for manipulation of targeted concepts, outperforming existing methods.

We show that reinforcement learning applied to large language models (LLMs) significantly boosts performance on complex coding and reasoning tasks. Additionally, we compare two general-purpose reasoning models - OpenAI o1 and an early checkpoint of o3 - with a domain-specific system, o1-ioi, which uses hand-engineered inference strategies designed for competing in the 2024 International Olympiad in Informatics (IOI). We competed live at IOI 2024 with o1-ioi and, using hand-crafted test-time strategies, placed in the 49th percentile. Under relaxed competition constraints, o1-ioi achieved a gold medal. However, when evaluating later models such as o3, we find that o3 achieves gold without hand-crafted domain-specific strategies or relaxed constraints. Our findings show that although specialized pipelines such as o1-ioi yield solid improvements, the scaled-up, general-purpose o3 model surpasses those results without relying on hand-crafted inference heuristics. Notably, o3 achieves a gold medal at the 2024 IOI and obtains a Codeforces rating on par with elite human competitors. Overall, these results indicate that scaling general-purpose reinforcement learning, rather than relying on domain-specific techniques, offers a robust path toward state-of-the-art AI in reasoning domains, such as competitive programming.

Test-time scaling is a promising new approach to language modeling that uses extra test-time compute to improve performance. Recently, OpenAI's o1 model showed this capability but did not publicly share its methodology, leading to many replication efforts. We seek the simplest approach to achieve test-time scaling and strong reasoning performance. First, we curate a small dataset s1K of 1,000 questions paired with reasoning traces relying on three criteria we validate through ablations: difficulty, diversity, and quality. Second, we develop budget forcing to control test-time compute by forcefully terminating the model's thinking process or lengthening it by appending "Wait" multiple times to the model's generation when it tries to end. This can lead the model to double-check its answer, often fixing incorrect reasoning steps. After supervised finetuning the Qwen2.5-32B-Instruct language model on s1K and equipping it with budget forcing, our model s1-32B exceeds o1-preview on competition math questions by up to 27% (MATH and AIME24). Further, scaling s1-32B with budget forcing allows extrapolating beyond its performance without test-time intervention: from 50% to 57% on AIME24. Our model, data, and code are open-source at https://github.com/simplescaling/s1