3 ms·
Ideally they would use the same prompt for LLMs as they use for Humans; and judge them in exactly the same manner. Only then is it a fair comparison.
by 1024core 3y ago
Ideally they would use the same prompt for LLMs as they use for Humans; and judge them in exactly the same manner. Only then is it a fair comparison.
- arjvik 3y agoWhy can’t the LLM be a (model, prompt prefix) pair?