In this ipython/jupyter tutorial I use Keras/Theano and an LSTM to predict sentiment for the IMDB dataset. This method is than compared empirically to TF-IDF and logistic regression. Finally, I pull the learned embeddings out of the model and analyze their structure with cosine search and T-SNE to demostrate what the LSTM has learned.
shubhampachori12110095 / imdb-lstm-tutorial Goto Github PK
View Code? Open in Web Editor NEWThis project forked from thenomemac/imdb-lstm-tutorial
In this tutorial I use an LSTM and Keras/Theano to predict sentiment on the IMDB dataset.