Simple extension of micrograd to consider a recurrent neural network (RNN) architecture. The RNN definition is found in extension.py. With collaboration of @amartorell98.
The only modification made to the original code is the inclusion of the extension.py
script. Within the playground.ipynb
file, you can locate the functions responsible for calculating the BPTT (Backpropagation Through Time) loss and conducting the training process. The data used for training is generated artificially. By adjusting parameters such as the number of epochs, gradient clipping value, learning rate value, and its decay, you can fine-tune the training process.