Hi everyone! I'm working on a security-focused system based on Orca architecture. The goal is to provide evidence-based responses that are both fast and reasonable. I'm specifically looking for advice on the best way to implement the reasoning traces and how to integrate a reliable knowledge base for the evidence part (possibly via RAG). Does anyone have experience with this approach, or can you point me toward any public projects or resources that use Orca-style distillation for specialized security tasks? Thanks!