Micronet is a library which gives you the ability to architect any permutations of sequential neural nets. It uses numpy
only.
# Example to build your own model
model = Sequential(
[
Linear(input_size, hidden_size_1),
Relu(),
Linear(hidden_size_1, hidden_size_2),
Relu(),
Linear(hidden_size_2, output_size),
Softmax(),
],
X = X,
Y = Y,
batch_size = 128,
loss_function = CrossEntropy()
)
model.train(epochs, alpha)