LSTMs, GRUs, and Advanced RNNs

Advanced models and variations of the plain Recurrent Neural Network (RNN) that overcome some RNNs practical drawbacks.

These notebooks serve us to practice some DL techniques, and as snippets to build on.

• Sentiment classification with RNN

• LSTM-Based sentiment classification model

• GRU-Based sentiment classification model

• Bi-directional LSTM-based sentiment classification model

• Stacked LSTM-based sentiment classification model

Training results comparison

Model	Epochs	Training time	Test accuracy
RNN	10	Low	0.83
LTSM	5	High	0.87
GRU	4	Medium	0.87
Bi-directional LSTM	4	Very High	0.88
Stacked	4	Very High	0.87