Add llm-as-judge preliminary results (too much non-evaluable)

Add llm-as-judge preliminary results (too much non-evaluable) #316