Is the tradeoff of the parallel output that we don't get arbitrary string generation? like output # of tokens is fixed ahead of time?
Either way, really cool and impressive.
Is the tradeoff of the parallel output that we don't get arbitrary string generation? like output # of tokens is fixed ahead of time?
Either way, really cool and impressive.
Yeah, it doesn't output strings, just decisions/answers.
Non-hallucinated ones at that.