Python: LSTM

Long Short Term Memory is primarily used for prediction tasks. LSTM was designed to be an improvement over traditional RNN models that only have one hidden state that is passed through. LSTM adds a layer of memory. This addition was able to bring significant advancements in language translation, speech recognition, and time series forecasting.
The training model was conducted using 1000 epochs. The first epoch had a loss of .2316 and the final epoch was reduced to .0080, though the lowest loss was at epoch 998, which was .0021. The graph of the loss results show a steep drop through the 600th epoch, where the results then flatten out.