in my experience limiting LLMs to a YES or NO gives very shaky results, ones that you could rely on maybe for something less critical like this but not ideal. I have used YES/NO for questioning if an LLM should reply or not and it has given less than stellar results even with the best models
26.11.2024 05:05
๐ 2
๐ 0
๐ฌ 1
๐ 0