First assignment in Deep Learning course - using NumPy
In this assignment I implemented a log-linear classifier and a multi-layer perceptron classifiers (one that has one hidden layer and another one which can handle an arbitrary number of layers). I also wrote the training code for those classifiers. All the classifiers use the hard cross-entropy loss. I tested the classifiers on the language identification task (both bigrams and unigrams representation), and the XOR task.