Kind of, ChatBot Arena was updated on Dec 7th, so hard to know for sure if Bard is just fine tuned against it. That being said Chatbot Arena is a pretty wide v

Kind of, ChatBot Arena was updated on Dec 7th, so hard to know for sure if Bard is just fine tuned against it. That being said Chatbot Arena is a pretty wide variety of test scenarios. If fine tuning made the model perfect, all the small models would get similar scores to GPT4, which they don't. Essentially it ranks how people believe a ChatBot should respond, rather than just zero shot, 1 shot and COT type benchmarks.