AllenNLP
it's better to learn the representation of all the modalities at the same timehttps://piazza.com/cmu/fall2018/11777/resources
32x64 elements in hm
32 from text representation
64 from image representation
(neglect some slides here)
AE
VAE
fig 1 and 2smooth at local scale but
KL Div tells the different between fig 1 and 2
网友评论