NeuroForge Documentation

🛠 Installation

NeuroForge is available via standard package managers and supports Python 3.9+. The framework includes GPU support through CUDA/OpenCL and quantum integration via Qiskit.

PIP Install

pip install neuroforge

For CUDA enabled systems. Automatically downloads cuDNN binaries.

Build from Source


git clone https://github.com/neuroforge/core.git
cd core
python setup.py install

Requires Python 3.9+, CUDA Toolkit, and CMake.

Documentation Version: 1.2.5 (2025-10-04)

🚀 Quick Start

Minimal Model

import neuroforge as nf

# Create a simple neural network
model = nf.Sequential([
    nf.Linear(128, 256),
    nf.Activation('GELU'),
    nf.Dropout(0.2),
    nf.QuantumLayer(256, 128),  # Quantum-integrated layer
    nf.Dense(128, 10)
])

# Compile with optimizer
model.compile(optimizer='qadam', loss='crossentropy')

# Dummy data
X = nf.random.normal(size=(1000, 128))
y = nf.random.integers(0, 9, size=(1000,))

# Train for 5 epochs
history = model.fit(X, y, epochs=5, batch_size=32)

This example demonstrates a basic hybrid-classical/quantum neural network using NeuroForge's API. The framework automatically handles GPU acceleration and quantum compilation when available.

⚡

Dynamic Training

Support for real-time model modification during training with automatic gradient tracking.

🤖

Quantum Integration

Hybrid quantum-classical layers with built-in quantum circuit compilation.

🌐

Distributed Training

Out-of-the-box support for multi-GPU and TPU distributed training operations.

🔐 API Reference

Core Classes

neuroforge.Sequential - Linear stack of layers
neuroforge.Model - Base class for custom models
neuroforge.QLayer - Quantum computing integration
neuroforge.Optimizer - Optimization framework
neuroforge.Callback - Training life-cycle management

Key Methods

model.compile()

Configure training model architecture

model.fit()

Train model with given data

model.evaluate()

Evaluate model performance

model.quantum()

Enable quantum compilation

model.save()

Serialize model to file

🧪 Code Examples

Quantum Neural Network

qmodel = nf.QuantumModel(
    n_qubits=4,  # Required quantum resources
    backend='qiskit',  # Quantum execution framework
    layers=[
        nf.QuantumLayer(4, 8, 'ry'),