Who's going to use Llama-3.1-405B to generate the best training dataset for small models?

→ View original post on X — @jiquanngiam