A tensorflow implementation of siamese lstm.
You can use this model to generate sentence-level embeddings so that you can calculate the similarity between two sentences.
Requirement: Tensorflow version= 1.1
Python verison= 3.5
pre-trained word2vec model
(GoogleNews-vectors-negative300.bin.gz)
from https://code.google.com/archive/p/word2vec/
2017-5-24 update:change dtype from tf.float32 to tf.float64