arXiv:2602.10273v1 Announce Type: new Abstract: Many recent reasoning gains in large language models can be explained as distribution sharpening: biasing generation toward high-likelihood trajectories already supported by the pretrained model, rather than modifying its weights. A natural formalization is the sequence-level power distribution $pi_alpha(ymid x)propto p_theta(ymid x)^alpha$ ($alpha>1$), which concentrates mass on whole sequences instead of adjusting token-level temperature. Prior work shows that Metropolis–Hastings (MH) sampling from this distribution recovers strong reasoning performance, but at order-of-magnitude inference slowdowns. […]

Ver mais

Einstein argues that religion evolved from primitive fear to moral guidance, but its highest form is “cosmic religious feeling”—an awe at the universe’s sublime order, free from anthropomorphic gods or dogma. This same feeling, he contends, drives genuine scientific inquiry. Rather than opponents, science and religion share a common wellspring: wonder at existence itself.

Ver mais

Recent advances in large language models (LLMs) have enabled the development of multimodal medical AI. While models such as MedGemini achieve high accuracy on VQA tasks like USMLE MM, their performance on ECG based tasks remains limited, and some models, such as MedGemma, do not support ECG data at all. Interpreting ECGs is inherently challenging, and diagnostic accuracy can vary depending on the interpreter’s experience. Although echocardiography provides rich diagnostic information, it requires specialized equipment and personnel, limiting […]

Ver mais

It feels like there’s currently a massive elephant in the room when it comes to ML, and it’s specifically around the idea that gradient descent might be a dead end in terms of a method that gets us anywhere near solving continual learning, casual learning, and beyond. Almost every researcher, whether postdoc, or PhD I’ve talked to feels like current methods are flawed and that the field is missing some stroke of creative genius. I’ve been told multiple […]

Ver mais

arXiv:2602.10233v1 Announce Type: new Abstract: Recent advances in LLM-guided evolutionary computation, particularly AlphaEvolve, have demonstrated remarkable success in discovering novel mathematical constructions and solving challenging optimization problems. In this article, we present ImprovEvolve, a simple yet effective technique for enhancing LLM-based evolutionary approaches such as AlphaEvolve. Given an optimization problem, the standard approach is to evolve program code that, when executed, produces a solution close to the optimum. We propose an alternative program parameterization that maintains the ability […]

Ver mais

arXiv:2602.10153v1 Announce Type: new Abstract: The AI Control research agenda aims to develop control protocols: safety techniques that prevent untrusted AI systems from taking harmful actions during deployment. Because human oversight is expensive, one approach is trusted monitoring, where weaker, trusted models oversee stronger, untrusted models$unicode{x2013}$but this often fails when the untrusted model’s actions exceed the monitor’s comprehension. We introduce legibility protocols, which encourage the untrusted model to take actions that are easier for a monitor to evaluate. […]

Ver mais

arXiv:2601.16298v1 Announce Type: new Abstract: With the rise of decentralized finance, fiat-to-cryptocurrency exchange platforms have become popular entry points into the cryptocurrency ecosystem. However, these platforms frequently fail to ensure adequate privacy protection, as evidenced by real-world breaches that exposed personally identifiable information (PII) and crypto addresses. Such leaks enable adversaries to link real-world identities to cryptocurrency transactions, undermining the presumed anonymity of cryptocurrency use. We propose FC-GUARD, a privacy-preserving exchange system designed to preserve user anonymity without […]

Ver mais

We study how neural emulators of partial differential equation solution operators internalize physical symmetries by introducing an influence-based diagnostic that measures the propagation of parameter updates between symmetry-related states, defined as the metric-weighted overlap of loss gradients evaluated along group orbits. This quantity probes the local geometry of the learned loss landscape and goes beyond forward-pass equivariance tests by directly assessing whether learning dynamics couple physically equivalent configurations. Applying our diagnostic to autoregressive fluid flow emulators, we show […]

Ver mais