(Depends on PR #18 #17) RNN Decoder with attention #15
Conversation
aasseman
added
enhancement
|
Seems to work well enough for now to run Will try to add the translation task from the pytorch tutorial, as to compare apples to apples if my implementation is doing what it should. |
aasseman
changed the title
|
This pull request introduces 1 alert when merging df60fb4 into aed980d - view on LGTM.com new alerts:
Comment posted by LGTM.com |
aasseman
changed the title
|
This pull request introduces 1 alert when merging 0138ef2 into aed980d - view on LGTM.com new alerts:
Comment posted by LGTM.com |
aasseman
changed the title
|
Ready for review. |
…/attn-decoder-rnn
tkornuta-ibm
left a comment
tkornuta-ibm
left a comment
There was a problem hiding this comment.
Please cleanup that wikitext_language_modeling_seq2seq.yml file
Aside of that, some of the changes were already merged... why GH is not seeing that? Are you squashing the commits or sth?
| hidden_size: 50 | ||
| num_layers: 1 | ||
| use_logsoftmax: False | ||
| <<<<<<< Updated upstream |
There was a problem hiding this comment.
Hmmmm, I guess the "updated upstream" is the right one... but not sure..
There was a problem hiding this comment.
I guess I got mixed up with my stashes. I checked-out the develop version of the two files in question, and that solved it.
2 participants
(Depends on PR #18 #17)
Adding RNN Decoder w/ attention, as per this pytorch tutorial. Itself based on the paper Neural Machine Translation by Jointly Learning to Align and Translate.