Juice Examples
CLI for running juice examples. More examples and benchmark tests can be found at the juice examples directory.
DISCLAIMER: Currently both CUDA and cuDNN are required for the examples to build.
Compile and call the build.
# install rust, if you need to
curl -sSf https://static.rust-lang.org/rustup.sh | sh
# download the code
git clone git@github.com:spearow/juice-examples.git && cd juice-examples
# build the binary
cargo build --release
# and you should see the CLI help page
../target/release/juice-examples --help
# which means, you can run the examples from belowNote for OSX El Capitan users: openssl no longer ships with OSX by default. brew link --force openssl should fix the problem. If not, see this Github issue for more details.
The MNIST Datasets comes not shipped with this repository (it's too big), but you can load it directly via the CLI.
# download the MNIST dataset.
../target/release/juice-examples load-dataset mnist
# run the MNIST linear example
../target/release/juice-examples mnist linear --batch-size 10
# run the MNIST MLP (Multilayer Perceptron) example
../target/release/juice-examples mnist mlp --batch-size 5 --learning-rate 0.001
# run the MNIST Convolutional Neural Network example
../target/release/juice-examples mnist conv --batch-size 10 --learning-rate 0.002The fashion-MNIST dataset is also too big to be included, but it can be downloaded in the same way as MNIST:
# download the fashion-MNIST dataset.
../target/release/juice-examples load-dataset fashion
# run the fashion-MNIST linear example
../target/release/juice-examples fashion linear --batch-size 10
# run the fashion-MNIST MLP (Multilayer Perceptron) example
../target/release/juice-examples fashion mlp --batch-size 5 --learning-rate 0.001
# run the fashion-MNIST Convolutional Neural Network example
../target/release/juice-examples fashion conv --batch-size 10 --learning-rate 0.002