DSReg: Provably Recovering Individual World Latents without Reconstruction Paper • 2610.09457 • Published 20 days ago • 3
Send a SCOUT First: Pre-hoc Reasoning for Adaptive Detector Allocation in Prompt-Injection Defense Paper • 2605.30837 • Published May 29 • 1
Send a SCOUT First: Pre-hoc Reasoning for Adaptive Detector Allocation in Prompt-Injection Defense Paper • 2605.30837 • Published May 29 • 1
Models Under SCOPE: Scalable and Controllable Routing via Pre-hoc Reasoning Paper • 2601.22323 • Published Jan 29 • 2
Position: Mechanistic Interpretability Should Prioritize Feature Consistency in SAEs Paper • 2505.20254 • Published May 26, 2025 • 5