Re:30% Drop In o1-Preview Accuracy When Putnam Problems Are ...
by amluto on Wednesday January 01, 2025 @03:02PM
(#42568808)
It seems more like ChatGPT was asked a rather bizarre question with far too little detail to make sense, and ChatGPT failed to notice or to ask for more information. Although it did get rather impressively confused about the pressure of the air.
[ Parent
| on HN ]
Re:30% Drop In o1-Preview Accuracy When Putnam Problems Are ...
by F7F7F7 on Wednesday January 01, 2025 @04:50PM
(#42569606)
“Failed to notice” is a hilariously weird way of talking about tokenization.
Re:30% Drop In o1-Preview Accuracy When Putnam Problems Are ...
by amluto on Wednesday January 01, 2025 @09:27PM
(#42571077)
Tokenization?
I mean that ChatGPT had no questions about the container of the gas (does it have weight? is it tared out?) or about buoyancy. And it’s really rather sad that ChatGPT calculated the volume of air at atmospheric temperature and pressure and didn’t notice that it was nowhere near fitting into its supposed container. (At least 1.01lb of liquid nitrogen would fit.)