Sequence to sequence learning for the same length output with LSTM

In this recipe, we will learn how to use LSTM to predict a value that is of the same or a slightly different length, such as subtraction of two numbers. 

Getting ready

Create a requirements.txt with Keras and six.moves dependencies. Import the relevant classes from keras, numpy, and six.moves as follows:

from __future__ import print_function
from keras.models import Sequential
from keras import layers
import numpy as np
import six.moves

In the next section, we will learn how to implement an LSTM network that can handle any three-digit subtraction.

How to do it…

1. Create a character table that can handle encoding and decoding. This class has three methods, as follows:
   • __init__()
   • encode()
   • decode()

2. The code is as follows:

class CharTable(object):
    def __init__(self, char):
        self.char = sorted(set(char))
        self.char_indices = dict((ch, i) for i, ch in enumerate(self.char))
        self.indices_char = dict((i, ch) for i, ch...