3 ms·
> we should call them "Text Inference Models”. “Predictive Text 2.0” seems about right, and also has the benefit of mapping onto a technology that most people
by Shish2k 4y ago
> we should call them "Text Inference Models”.
“Predictive Text 2.0” seems about right, and also has the benefit of mapping onto a technology that most people are already familiar with
- thomastjeffery 4y agoI think predictive is a bit of a stretch, too. There isn't any prediction happening. As complex as they can be, every continuation is wholly dependent on the text in the training corpus and the text in the prompt.