Deploying 1-bit Bonsai-27B via PrismML llama.cpp fork

The tutorial walks through serving the 1-bit Bonsai-27B locally with the PrismML llama.cpp fork, which adds the CUDA kernels required to decode its Q1_0_g128 GGUF quantization format. It covers GPU runtime validation, Python dependency setup, and OpenAI-compatible local inference workflows.