RealCompanion: Benchmarking Human Understanding from Reasoning over Longitudinal Real-World Conversations Paper • 2610.01780 • Published Oct 1 • 40
WorldLine: Action-Driven Visual Simulation for Robotic Manipulation Paper • 2609.38059 • Published Sep 29 • 22
TabFM-Auto: Self-Evolving Pipelines for Tabular Foundation Models Paper • 2609.37989 • Published Sep 29 • 12
Understanding On-Policy Distillation: A Mechanistic Interpretability Perspective via Sparse Crosscoders Paper • 2609.35210 • Published Sep 28 • 4
PreviewDiff: Multimodal Critic-Guided Search over Diffusion Latents Paper • 2609.36199 • Published Sep 28 • 6