arXiv:2603.00675v1 Announce Type: new Abstract: Foundation models pre-trained on large-scale datasets demonstrate strong transfer learning capabilities; however, their adaptation to complex multi-label diagnostic tasks-such as comprehensive head CT finding detection-remains understudied. Standard parameter-efficient fine-tuning methods such as LoRA apply uniform adaptations across pathology types, which may limit performance for diverse medical findings. We propose a Mixture of Low-Rank Experts (MoLRE) framework that extends LoRA with multiple specialized low-rank adapters and unsupervised soft routing. This approach enables conditional feature […]

Ver mais

arXiv:2602.11164v1 Announce Type: new Abstract: Automated optimization modeling via Large Language Models (LLMs) has emerged as a promising approach to assist complex human decision-making. While post-training has become a pivotal technique to enhance LLMs’ capabilities in this domain, its effectiveness is severely constrained by the scarcity and underutilization of high-quality training data. However, through a detailed profiling of error patterns across various problem-response pairs drawn from post-training, we identify two fundamental limitations of existing automated optimization modeling approaches: […]

Ver mais

In this tutorial, we explore Microsoft VibeVoice in Colab and build a complete hands-on workflow for both speech recognition and real-time speech synthesis. We set up the environment from scratch, install the required dependencies, verify support for the latest VibeVoice models, and then walk through advanced capabilities such as speaker-aware transcription, context-guided ASR, batch audio processing, expressive text-to-speech generation, and an end-to-end speech-to-speech pipeline. As we work through the tutorial, we interact with practical examples, test different voice […]

Ver mais

Congrats to the @cursor_ai team on the launch of Composer 2! We are proud to see Kimi-k2.5 provide the foundation. Seeing our model integrated effectively through Cursor’s continued pretraining & high-compute RL training is the open model ecosystem we love to support. Note: Cursor accesses Kimi-k2.5 via @FireworksAI_HQ hosted RL and inference platform as part of an authorized commercial partnership. — Kimi.ai @Kimi_Moonshot, responding to reports that Composer 2 was built on top of Kimi K2.5 Tags: kimi, […]

Ver mais

By Michael Nielsen Attention conservation notice: Rough notes, very rapidly written, by an interested non-expert outsider. Rich Sutton is an expert on neural networks at the University of Alberta and DeepMind. He’s written a stimulating essay describing what he calls “the bitter lesson”: in AI research it’s extremely seductive to try to build expert domain knowledge into the systems you’re creating, but, according to Sutton, this approach gets beaten again and again by methods leveraging brute force computation, […]

Ver mais

Hi all, I’m training a quadruped robot (Isaac Gym / legged_gym style) and trying to achieve a policy that switches between: – command = 0 → stable quadruped standing – command = 1 → stable bipedal standing (hind legs only) However, I’m facing several issues that seem related to reward scaling and interference between reward terms. Current reward components: – zero linear/angular velocity tracking – projected gravity alignment – quadruped base height reward – bipedal base height reward […]

Ver mais

arXiv:2602.17846v1 Announce Type: new Abstract: Diffusion models generate high-quality samples but can also memorize training data, raising serious privacy concerns. Understanding the mechanisms governing when memorization versus generalization occurs remains an active area of research. In particular, it is unclear where along the noise schedule memorization is induced, how data geometry influences it, and how phenomena at different noise scales interact. We introduce a geometric framework that partitions the noise schedule into three regimes based on the coverage […]

Ver mais

arXiv:2602.15900v1 Announce Type: new Abstract: Robot perception under low light or high dynamic range is usually improved downstream – via more robust feature extraction, image enhancement, or closed-loop exposure control. However, all of these approaches are limited by the image captured these conditions. An alternate approach is to utilize a programmable onboard light that adds to ambient illumination and improves captured images. However, it is not straightforward to predict its impact on image formation. Illumination interacts nonlinearly with […]

Ver mais

arXiv:2603.02452v1 Announce Type: cross Abstract: A major focus in designing methods for learning distributions defined on manifolds is to alleviate the need to implicitly learn the manifold so that learning can concentrate on the data distribution within the manifold. However, accomplishing this often leads to compute-intensive solutions. In this work, we propose a simple modification to denoising score-matching in the ambient space to implicitly account for the manifold, thereby reducing the burden of learning the manifold while maintaining […]

Ver mais

arXiv:2604.09838v1 Announce Type: new Abstract: We present a novel diffusion-based framework for synthesizing 2D vector fields from sparse, coherent inputs (i.e., streamlines) while maintaining physical plausibility. Our method employs a conditional denoising diffusion probabilistic model with classifier-free guidance, enabling progressive reconstruction that preserves both geometric and physical constraints. Experimental results demonstrate our method’s ability to synthesize plausible vector fields that adhere to physical laws while maintaining fidelity to sparse input observations, outperforming traditional optimization-based approaches in terms of […]

Ver mais