bertmoons / quantizedneuralnetworks-lasagne-theano

Training Quantized Neural Networks

Introduction

Train your own Quantized Neural Networks (QNN) - networks trained with quantized weights and activations - in Lasagne / Theano. If you use this code, please cite "B.Moons et al. "Minimum Energy Quantized Neural Networks", Asilomar Conference on Signals, Systems and Computers, 2017". Take a look at our presentation or at the paper on arxiv.

This code is based on a lasagne/theano and a Keras/Tensorflow version of BinaryNet.

Preliminaries

Running this code requires:

1. Theano
2. Lasagne
3. pylearn2 + the correct PYLEARN2_DATA_PATH in your shell
4. A GPU with recent versions of CUDA and CUDNN
5. Correct paths in the top of train_quantized.py

Training your own QNN

This repo includes toy examples for CIFAR-10 and MNIST. Training can be done directly from the command line:

python train_quantized.py

The following parameters are crucial:

• -ds: dataset, either 'CIFAR-10' or 'MNIST'
• -nb: number of bits for weights and activations
• -q: if 1, the networks is quantized, otherwise it is floating point
• -ne: number of epochs, 100 should do
• -ft: if 1, a previous model is used to finetune, otherwise training is from scratch
• -nl: the type of nonlinearity required, can either be 'relu' or 'hardtanh'
• -sym: if nl=relu, should be 0, otherwise it should be 1
• -lr: learning rate, 0.01 or 0.001 should be fine
• nl<>: the number of layers in block A, B, C
• nf<>: the number of filters in block A, B, C

Examples

The included networks have parametrized sizes and are split into three blocks (A-B-C), each with a number of layers (nl) and a number of filters per layer (nf).

• This is how to train a 4-bit full qnn on CIFAR-10:

  python train_quantized.py -ds CIFAR-10 -nb 4 -q 1 -ne 100 -nla 1 -nfa 64 -nlb 1 -nfb 64 -nlc 1 -nfc 64 -ft 0 -nl relu -sym 0 -lr 0.001

• This is how to train a BinaryNet on CIFAR-10:

  python train_quantized.py -ds CIFAR-10 -nb 0 -q 1 -ne 100 -nla 1 -nfa 64 -nlb 1 -nfb 64 -nlc 1 -nfc 64 -ft 0 -nl hardtanh -sym 1 -lr 0.001