The model which is created is used to convert human readable dates like "April 20 1999" to "1999-04-20". It was created using Attention mechanism wherein the output depends only on certain regimes of the input to generate a character of an ouput. The dates were generated using the faker module, where we run through a loop to get the required dataset. The characters of both the dates(human readable and machine readable) were mapped to integers in a dictionary which was further used to preprocess the dataset where we convert the dataset to a one-hot encoding This model was created using the Attention mechanism with the pre-activation being a bidirectional LSTM and the post activation layer being unidirectional LSTM. The weights alpha(the attention paid at timestep t to generate the output characer was calculated using activations from both pre-attention and post-attention. The attention can be seen in the generated plot in the code.