Practical-pytorch: Bahdanau Decoder Implementation
Hi @spro,
Thanks for really great explanation of decoder, especially for Bahdanau decoder. But, i'm little bit confuse about code in __init__ function of BahdanauAttnDecoderRNN class.
self.attn = GeneralAttn(hidden_size)
I can't find any class that define GeneralAttn. This is built-in class? Can you please elaborate for this? Thanks again!
sakinaljana
👍1
All 6 comments
Good catch, it was originally split out as 3 separate attention modules (GeneralAttn, DotAttn, ConcatAttn) instead of one with an argument to choose the strategy. Further, they actually used the "concat" strategy. So this should be self.attn = Attn("concat", hidden_size)
spro
on 23 May 2017
👍5
Cool, Thanks for the clarification!
sakinaljana
on 23 May 2017
Can you please change that line on the notebook?
rafaelvalle
on 20 Nov 2017
👍4
Still not changed. Hope somebody could do it.
poweihuang17
on 15 Aug 2018
👍1
119 fixes this and some more issues with Bahdanau decoder.
anantzoid
on 30 Oct 2018
@anantzoid it's still not fixes in tutorial.
kyquang97
on 8 May 2019
Was this page helpful?
0 / 5 - 0 ratings
Related issues
spro
路
8Comments
AuCson
路
3Comments
EinAeffchen
路
5Comments
kdrivas
路
3Comments
rk2900
路
6Comments
Most helpful comment
Good catch, it was originally split out as 3 separate attention modules (GeneralAttn, DotAttn, ConcatAttn) instead of one with an argument to choose the strategy. Further, they actually used the "concat" strategy. So this should be
self.attn = Attn("concat", hidden_size)