Sovereign AI for Government DevOps: Replication Package

Replication package for "Sovereign AI for Government DevOps: Benchmarking Quantized LLMs on Commodity Hardware". Contains:- Performance benchmark data (latency, throughput, accuracy)- Energy consumption measurements- Cost analysis across utilisation scenarios- Log analysis evaluation results (HDFS, BGL datasets)- Infrastructure-as-Code generation results with failure breakdown- Root cause analysis human evaluation scores- Calibration datasets for imatrix quantization- System prompts, random seeds, and sampling parameters- Python scripts for reproducing paper tables and statistical analysis Hardware configurations:- Cloud: AWS g5.12xlarge (4× NVIDIA A10G)- Sovereign: MacBook Pro M2 (32GB) Model: Qwen3-32B (FP16 cloud, Q4_K_M sovereign)