In this notebook, I tested out an architecture inferred from research done on the best LSTM hyperparameters for sequence classification. This architecture performed surprisingly poorly, and it is evident that far more work is needed to bring us over the 70% plateau.
In this notebook, I tested out an architecture inferred from research done on the best LSTM hyperparameters for sequence classification. This architecture performed surprisingly poorly, and it is evident that far more work is needed to bring us over the 70% plateau.