lunny/keras
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix misleading comment in babi_memnn.py

Browse Source 
Output of question_encoder is 3D : sequence of vectors, not single vector.
This commit is contained in:
Fariz Rahman 2015-11-23 03:54:32 +05:30
parent 05aecbc0bc
commit d322785543
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
examples/babi_memnn.py
View File

@ -155,7 +155,7 @@ input_encoder_m.add(Embedding(input_dim=vocab_size,
output_dim=64, output_dim=64,
input_length=story_maxlen)) input_length=story_maxlen))
# output: (samples, story_maxlen, embedding_dim) # output: (samples, story_maxlen, embedding_dim)
# embed the question into a single vector # embed the question into a sequence of vectors
question_encoder = Sequential() question_encoder = Sequential()
question_encoder.add(Embedding(input_dim=vocab_size, question_encoder.add(Embedding(input_dim=vocab_size,
output_dim=64, output_dim=64,