Partially agree, but it's a continuous value rather than a boolean. We've seen LLM performance largely follow this story: https://twitter.com/karpathy/status/16

Partially agree, but it's a continuous value rather than a boolean. We've seen LLM performance largely follow this story: https://twitter.com/karpathy/status/1655994367033884672/photo/1 https://twitter.com/karpathy/status/1655994367033884672/phot... From benchmarking, we've been positively surprised by how effective few-shot learning and PEFT are, at closing the domain gap. "When it encounters novel data (value) it will likely perform poorly" -- is that not true of human annotators too? :)