3 ms·
The text returned by the tool itself makes it not "next token prediction". Aside from having side effects, the reason it's helpful is that it's out of distribut
by astrange 1y ago
The text returned by the tool itself makes it not "next token prediction". Aside from having side effects, the reason it's helpful is that it's out of distribution for the model. So it changes the properties of the system.
- porridgeraisin 1y agoAh ok, understood what you meant.
- SEGyges 1y agoThis is true of the system as a whole, but the core neural network is still a next-token predictor.