Is your network only implement encoder-decoder ?

Question

Is your network only implement encoder-decoder ?

Closed this issue 4 years ago · 2 comments

There is two stage in the paper, but i find the DIMModel only implement stage one. And the loss only include alpha_prediction loss, there is no composition loss. Is that true?

foamliu commented 4 years ago

Yes.

Answer 1 · 2020-06-16T10:06:35.000Z

Same question