3 ms·
RAG does not guarantee elimination of hallucination it seems. If the foundation model's training outweighs what it is in the external source, it will still hall
by wenc 3y ago
RAG does not guarantee elimination of hallucination it seems. If the foundation model's training outweighs what it is in the external source, it will still hallucinate.
Interestingly ChatGPT-4 gets it right. I tried your first question with ChatGPT-4:
User: what’s the single-engine service ceiling of a Baron 55
ChatGPT: The single-engine service ceiling of a Beechcraft Baron 55 is approximately 7,000 feet. This is the maximum altitude at which the aircraft can maintain a specified rate of climb, usually 100 feet per minute, with one engine inoperative.
- brandall10 3y agoA common approach with RAG is to tell it to only used the supplied data. I have yet to see that not work using 3.5, though it doesn’t mean it won’t hallucinate if the injected text does not have the answer to the question.