TorontoDeepLearning / convnet

A GPU implementation of Convolutional Neural Nets in C++

Welcome to ConvNet.

ConvNet is a fast C++ based GPU implementation of Convolutional Neural Nets.

• Supports Multi-GPU architectures (Multiple GPUs, Single machine).
• Provides a fast CPU-only feature extractor.

Installation

[Install guide] (https://github.com/torontodeeplearning/convnet/blob/master/INSTALL)

Pre-trained Models

Pre-trained models and examples for training and feature extraction are provided for

• Imagenet Classification (ILSVRC 2013)
• MNIST, Feed-forward Neural Net
• MNIST, Convolutional Neural Net

Tutorials

Coming soon.

Documentation

here