Hopefully you found the answer already. But in case you didn't or someone else have the same questions, it can be fine like this.
x[:,:,y] This is called advanced indexing. Theano and NumPy support this. Fred Le 22 déc. 2016 10:17, "Quanty" <[email protected]> a écrit : > Specifically, I have tensor x of shape (n1,n2,n3) and matrix y of shape > (n1,n2). Each element in y is a iscalar between (0,n3-1). I hope to choose > the element in x using y as the index, so that I can get a matrix of > (n1,n2). > > -- > > --- > You received this message because you are subscribed to the Google Groups > "theano-users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. > -- --- You received this message because you are subscribed to the Google Groups "theano-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
