cap2vid icon indicating copy to clipboard operation
cap2vid copied to clipboard
verified publisher icon Singularity42

How are words in the captions represented in the BiLSTM model?

Open Namrata96 opened this issue 7 years ago • 0 comments

They are mapped to unique integers while saving the captions in h5py file. However, before feeding them into the bidirectional are they represented as an embedding/fixed size vector? or the y_encode function in this comment does it?

Namrata96 avatar Apr 08 '18 20:04 Namrata96