Test more chat models and compare them to Llama 2

[gtabot]

As a developer I want to create a methodology that can consistently compare and contrast LLM models So that we can systematically benchmark models