i’m thinking of self-attention a little during a funny time

the formula from wikipedia is softmax(QK^T)V

this to me leaves it a little ambiguous how the dimension broadcasting works :s

1020 blarf 1021 :s

1107

traffick boss is deflated

he lays on the ground, fluids and large organs seemingly missing, bones collapsed

his hands, feet, and some of his head remain with size 1108