3 ms·
When the answer is something ridiculous or stupid, it's 95%+ of the time GPT-3.5-turbo and is rarely disclosed by the other party. GPT-4 is an order of magnitud
by icelancer 3y ago
When the answer is something ridiculous or stupid, it's 95%+ of the time GPT-3.5-turbo and is rarely disclosed by the other party. GPT-4 is an order of magnitude better, if not two orders of magnitude better.
It's hard to tell if the party crapping on ChatGPT is doing so out of ignorance or malice.
Finetuning with GPT-4 can't come soon enough...