We used Numeral Decomposer 1.1 for a Reinforcement Learning Algorithm.

In 'Code.py' there is the function learn_language, which lets you pass a language* and run the reinforcement learning of numerals. In 'Learned English', we have provided a detailed example of such a reinforcement learning. In 'How each language got learned', there is a shortened documentation of the reinforcement learning for each language.

*Pass a language by writing one the strings listed in 'Languages'. For some of the languages you also need to download 'Numeral.csv' and link its path to FILE_PATH_OF_LANGUAGESANDNUMBERS_DATA_CSV in the code.