AI研究論文每日精選

GuardReasoner: Towards Reasoning-based LLM Safeguards

Thoughts Are All Over the Place: On the Underthinking of o1-Like LLMs

Streaming DiLoCo with overlapping communication: Towards a Distributed Free Lunch

o3-mini vs DeepSeek-R1: Which One is Safer?

Large Language Models Think Too Fast To Explore Effectively

MedXpertQA: Benchmarking Expert-Level Medical Reasoning and Understanding

WILDCHAT-50M: A Deep Dive Into the Role of Synthetic Data in Post-Training

SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformer

PhysBench: Benchmarking and Enhancing Vision-Language Models for Physical World Understanding

CowPilot: A Framework for Autonomous and Human-Agent Collaborative Web Navigation