You're incorrect. > For example, LLaMA's 13B architecture outperforms GPT-3 despite being 10 times smaller. Of course this is all on benchmarks but it's a big

You're incorrect. > For example, LLaMA's 13B architecture outperforms GPT-3 despite being 10 times smaller. Of course this is all on benchmarks but it's a big improvement if true.