NeuralNet

A deep learning library from scratch using NumPy and SciPy.

Components

Component	Description
NeuralNet.Trainer	Has the Trainer used to train any neural network
NeuralNet.nn	Has the sequential class to combine layers
NeuralNet.layers	Has the implementation of various layers
NeuralNet.activations	Has the implementation of activation functions
NeuralNet.optim	Has the implementation of various optimizers
NeuralNet.loss	Has the implementation of various loss functions

Example

# Necessary imports
from NeuralNet.layers import Linear
from NeuralNet.activations import ReLU
import NeuralNet.optim as optim
from NeuralNet.losses import MSE
from NeuralNet import Trainer, Sequential

# Define the network
net = nn.Sequential(
    [
        Linear(1, 10),
        ReLU(),
        Linear(10, 1)
    ]
)
# Train the model
Trainer.train(net=net, inputs=inputs, targets=targets, loss_fn=MSE(), optimizer=optim.SGD(lr=1e-3))