AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Is There a Log Distribution of Time Cuonsense?
There's people who've argued, i'm not expert, and like auditory cortex, there's people arguing that there ought to be a log distribution of time cuonsense and auditory ortex. A, having log scale at every layer of the network ands a letting the model do the following thing. If i've trained on speech, ah, such a, you know, we do like auditory emnist, and the end of the modelas been trained to decode one or four or eight, we can give it arbitrarily. We can give it slowed speech or sped up speech in go sad then. And the network goes, oh, yes, that's seven, and it's really slow