Defines the model's role and behaviour.
Maximum tokens the model will generate.
Higher = more creative. Lower = more focused.
Number of top tokens considered at each step.