3 ms·
If all or most LLMs are susceptible to failing this test, why would an LLM be a good means of evaluating performance on this test without being given a rubric?
by binary132 3mo ago
If all or most LLMs are susceptible to failing this test, why would an LLM be a good means of evaluating performance on this test without being given a rubric?
- layer8 3mo agoBecause evaluating a performance is different task from creating a performance. Someone who plays an instrument badly often has a different perception from someone who has to listen to it. ;)