1. Insights from Verification: Training a Verilog Generation LLM with Reinforcement Learning with Testbench Feedback; 2. Circuitdiff: Bridging Netlist Knowledge with RTL Based on Graph Denoising Diffusion