NeuralPy is a High-Level Keras like deep learning library that works on top of PyTorch written in pure Python. NeuralPy can be used to develop state-of-the-art deep learning models in a few lines of code. It provides a Keras like simple yet powerful interface to build and train models.

Here are some highlights of NeuralPy

  • Provides an easy interface that is suitable for fast prototyping, learning, and research
  • Can run on both CPU and GPU
  • Works on top of PyTorch
  • Cross-Compatible with PyTorch models


PyTorch is an open-source machine learning framework that accelerates the path from research prototyping to production deployment developed by Facebook runs on both CPU and GPU.

According to Wikipedia,

PyTorch is an open-source machine learning library based on the Torch library, used for applications such as computer vision and natural language processing, primarily developed by Facebook's AI Research lab (FAIR). It is free and open-source software released under the Modified BSD license.

NeuralPy is a high-level library that works on top of PyTorch. As it works on top of PyTorch, NerualPy supports both CPU and GPU and can perform numerical operations very efficiently.

If you want to learn more about PyTorch, then please check the PyTorch documentation.


To install NeuralPy, open terminal window type the following command:

pip install neuralpy-torch

If you have multiple versions of it, then you might need to use pip3.

pip3 install neuralpy-torch
python3 -m pip install neuralpy-torch

NeuralPy requires Pytorch and Numpy, first install those

Check the documentation for Installation related information


The only dependencies of NeuralPy are Pytorch (used as backend) and Numpy.

Get Started

Let's create a linear regression model in 100 seconds.

Importing the dependencies

import numpy as np

from neuralpy.models import Sequential
from neuralpy.layers import Dense
from neuralpy.optimizer import Adam
from neuralpy.loss_functions import MSELoss

Making some random data

# Random seed for numpy

# Generating the data
X_train = np.random.rand(100, 1) * 10
y_train = X_train + 5 *np.random.rand(100, 1)

X_validation = np.random.rand(100, 1) * 10
y_validation = X_validation + 5 * np.random.rand(100, 1)

X_test = np.random.rand(10, 1) * 10
y_test = X_test + 5 * np.random.rand(10, 1)

Making the model

# Making the model
model = Sequential()
model.add(Dense(n_nodes=1, n_inputs=1, bias=True, name="Input Layer"))

# Compiling the model
model.compile(optimizer=Adam(), loss_function=MSELoss())

# Printing model summary

Training the model, y_train), test_data=(X_validation, y_validation), epochs=300, batch_size=4)

Predicting using the trained model

model.predict(X=X_test, batch_size=4)