Yelp Sentence Sentiment Analysis

install requirements

python >= 3.6.3

$ pip install requirements.txt

dataset

• used the 100k subset of the Yelp Dataset

training and validating models

AWD-LSTM

$ python main.py --total_epochs 20

SimpleLSTM

$ python main.py --model simple-lstm

BidirectionalLSTM

$ python main.py --model bidirectional-lstm

results so far

Model	Training Accuracy	Validation Accuracy
SimpleLSTM	70.88%	67.2%
BidirectionalLSTM	70.75%	68.56%
AWD-LSTM (no pretraining)	72.85%	69.37%