Storia: Improving Bash Generation in Small Language Models with Grammar-Constrained Decoding | NVIDIA Technical Blog — Warptech Lab News