planning to implement Attend It Again paper. about drqa HOT 2 CLOSED

It's hard to tell, but you can give it a try. The model design choice heavily depends on the dataset and it's hard to predict whether a method works in advance :)

from drqa.

Thanks for your reply!
I have a question

doc_hiddens is a tensor which is of shape:

batch_size X number_of_sequence X doc_hidden_size

for example: 10 X 253 X 768

The input x1_emb is of shape:

batch_size X number_of_sequence X embedding_size

for example: 10 X 253 X 300

Since dimensions of these 2 tensors doc_hiddens and x1_emb are different, What I can do is put the x1_emb into an LSTM and get this as the same dimension as of doc_hiddens.

Then I can do a point-wise addition between doc_hiddens and x1_emb to get an unified-weighted tensor.

Can you share if point-wise addition would be a good operation in this case? or there are other better operations?

Actually, I am a bit confused on how to combine these 2 tensors, can you share any idea?

from drqa.