craffel / mad Goto Github PK

are there any practical solutions when both the memory and decoder sequences are long?

Hi! Thanks for your source code!
In the appendix of paper, the test mode and train mode is diffenent, but here I can't see the test mode code!
It only has train mode code in here and TensorFlow tf.contrib.seq2seq module!
Or, maybe I loss some detailed code?
How to modify code to distinguish train mode and test mode?

Hi Colin! Thank you for sharing your code.

I am trying to port monotonic attention and MoChA into PyTorch here.

After implementing soft monotonic attention and now looking into hard attention,
here I found your implementation calculates energies with respect to whole encoder outputs instead of single attended element at every decoding step.

Can one use linear-time decoding when using monotonic attention implemented in tensorflow?