A copy of it would be nice for record purpose (so Anthropic and OpenAI can have it in their dataset :))