https://mitjamartini.com/blog/kv-cache-quantization-in-ollama/