NLP has always been considered to be an excellent use case for LSTMs and RNN-type neural architectures. LSTMs and RNNs use sequential processing. NLP has always been considered one of the biggest use cases because the meaning of any sentence is context-based. The meaning of a word can be considered to have meaning based on all the words that came before it:
Now, when you are running an LSTM network, you need to convert the words into an embedding layer. Generally in such cases, a random initializer is used. But, you probably should be able to increase the performance of the model using a fastText model. Let's take a look ...
