Reinforcement learning and classical planning are typically seen as two distinct problems, with differing formulations necessitating different solutions. Yet, when humans are given a task, regardless of the way it is specified, they can often derive the additional information needed to solve the problem efficiently. The key to this ability is introspection: by reasoning about their internal models of the problem, humans directly synthesize additional task-relevant information. In this paper, we propose that this introspection can be thought […]

Ver mais

arXiv:2602.23407v1 Announce Type: new Abstract: Large language models (LLMs) have demonstrated strong capabilities in code generation, yet they remain prone to producing security vulnerabilities. Existing approaches commonly suffer from two key limitations: the scarcity of high-quality security data and coarse-grained reinforcement learning reward signals. To address these challenges, we propose Vul2Safe, a new secure code generation framework that leverages LLM self-reflection to construct high-confidence repair pairs from real-world vulnerabilities, and further generates diverse implicit prompts to build the […]

Ver mais

This paper addresses the problem of training a reinforcement learning (RL) policy under partial observability by exploiting a privileged, anytime-feasible planner agent available exclusively during training. We formalize this as a Partially Observable Markov Decision Process (POMDP) in which a planner agent with access to an approximate dynamical model and privileged state information guides a learning agent that observes only a lossy projection of the true state. To realize this framework, we introduce an anytime-feasible Model Predictive Control […]

Ver mais

Agentic Engineering Patterns > LLMs are restricted by their context limit – how many tokens they can fit in their working memory at any given time. These values have not increased much over the past two years even as the LLMs themselves have seen dramatic improvements in their abilities – they generally top out at around 1,000,000, and benchmarks frequently report better quality results below 200,000. Carefully managing the context such that it fits within those limits is […]

Ver mais

Imagine this: you’re in a team meeting, and your manager tells you there’s an old service, written a couple of years ago, that used to work fine but now breaks in some weird way. And you need to fix it. On top of that, they ask you to implement a new feature that the team has been wanting for a long time. At this point, your happiness for the next few hours, days, or even weeks largely depends […]

Ver mais

Everyone else is telling you to learn more, read more, stack more frameworks, chase more insights. I’m here to say the opposite: the most powerful thing you can do this year is deliberately know less.  I’m not talking about becoming stupid. I’m talking about unlearning the borrowed certainty that’s keeping you stuck.  Because here’s what I’ve seen, in myself first, then in way too many people who message me: The smartest, most thoughtful people are often the most stalled. You’ve […]

Ver mais

arXiv:2604.14265v1 Announce Type: new Abstract: We study behavior-regularized reinforcement learning (RL), where regularization toward a reference distribution (the dataset in offline RL or the base model in LLM RL finetuning) is essential to prevent value over-optimization caused by erroneous out-of-distribution extrapolation. Existing methods either rely on reparameterized policy gradient, which are difficult to scale to large generative models, or on reject sampling, which can be overly conservative when attempting to move beyond the behavior support. In this paper, […]

Ver mais

arXiv:2601.00092v1 Announce Type: new Abstract: 4D spatial intelligence involves perceiving and processing how objects move or change over time. Humans naturally possess 4D spatial intelligence, supporting a broad spectrum of spatial reasoning abilities. To what extent can Multimodal Large Language Models (MLLMs) achieve human-level 4D spatial intelligence? In this work, we present Spatial4D-Bench, a versatile 4D spatial intelligence benchmark designed to comprehensively assess the 4D spatial reasoning abilities of MLLMs. Unlike existing spatial intelligence benchmarks that are often […]

Ver mais

arXiv:2603.10091v1 Announce Type: new Abstract: The widespread adoption of thinking mode in large language models (LLMs) has significantly enhanced complex task processing capabilities while introducing new security risks. When subjected to jailbreak attacks, the step-by-step reasoning process may cause models to generate more detailed harmful content. We observe that thinking mode exhibits unique vulnerabilities when processing interleaved multiple tasks. Based on this observation, we propose multi-stream perturbation attack, which generates superimposed interference by interweaving multiple task streams within […]

Ver mais