3 ms·
I thought we don’t believe models when they tell us their training cutoff? Isn’t self-description particularly subject to hallucination?
by didgeoridoo 2y ago
I thought we don’t believe models when they tell us their training cutoff? Isn’t self-description particularly subject to hallucination?
- Filligree 2y agoOpenAI puts the “knowledge cutoff” in the system prompt. It’s just repeating that.
- didgeoridoo 2y agoInteresting, mine says October 2023, which makes sense as it knows about APIs from WWDC2023 (although it needs a lot of prodding and correcting to get their usage right).
- sebastiennight 2y agoAsk GPT-4o for the cutoff date and it will likely answer "October 2023". And then ask the model when GPT-4-Turbo was released, and it will accurately reply November 6th, 2023. You can't use those models to ask about self-aware information, this is just asking them to hallucinate a plausible answer.