jordanspooner

#=================================
# Sentence-level QE -- BiRNN model
#=================================
#
## Inputs:
# 1. Sentences in src language (shape: (mini_batch_size, line_words))
# 2. Parallel machine-translated documents (shape: (mini_batch_size, line_words))
#
## Output:
# 1. Sentence quality scores (shape: (mini_batch_size,))