This repository contains a LeNet Caffe Model trained on the MNIST dataset, based on the instructions available at https://caffe.berkeleyvision.org/gathered/examples/mnist.html.
The MNIST test dataset is included in this repository, in the LMDB format. It
was originally downloaded from http://yann.lecun.com/exdb/mnist and converted
to the LMDB format using Caffe's convert_mnist_data tool.
This model was trained using:
caffe train -solver=lenet_solver.prototxtThe train dataset is not included in this repository.
Run this to test the model:
caffe test -model=lenet_train_test.prototxt -weights=lenet_iter_10000.caffemodelThe model trained for 10000 iterations has a test accuracy of 98.6% and a
softmax loss of 0.0444183.
The contents of this repository are released for unrestricted use.