CODEWITHSUNDEEP
keraslstmrecurrent-neural-network
176coding
176coding

Reputation: 3059

How to use keras to implement second many-to-many architecture of LSTM

I have searched on google, most blog is about many-to-one or the first many-to-many architecture, is there any example for the last architecture. thx

enter image description here

Upvotes: 0

Views: 638

Answers (1)

Songlin Liu
Songlin Liu

Reputation: 11

You can just set the timestep=input_size+output_size-1,and the last (output_size-1) input is all zeros;at last layer,add a masking layer to filter the first (input_size-1) output.So we can get a input_size to output_size LSTM architecture.

Upvotes: 1

Related Questions