Also, training over ChatGPT outputs is terribly expensive and Anthropic already has a working competing paradigm for fine-tuning (Claude worked before ChatGPT w

Also, training over ChatGPT outputs is terribly expensive and Anthropic already has a working competing paradigm for fine-tuning (Claude worked before ChatGPT was released).