Prism-Δ: Differential Subspace Steering for Prompt Highlighting in Large Language Models Paper • 2603.10705 • Published 1 day ago • 10
Injecting External Knowledge into the Reasoning Process Enhances Retrieval-Augmented Generation Paper • 2507.19333 • Published Jul 25, 2025 • 1
Annotation-Efficient Universal Honesty Alignment Collection Official Collections of paper "Annotation-Efficient Universal Honesty Alignment". • 5 items • Updated Oct 21, 2025 • 3