3 ms·
From my naïve point of view it seems obvious that at any point where both an „a“ or a „an“ would fit the model randomly selects one of them and by doing so redu
by laszlokorte 4y ago
From my naïve point of view it seems obvious that at any point where both an „a“ or a „an“ would fit the model randomly selects one of them and by doing so reduces the set of possible nouns to follow.
- Sharlin 4y agoNo, of course not randomly. If you RTA, it should be clear that some prompts are more likely to be followed by an “an” than others.