Apply filter in docker container, supporting GPU

[matthieudelaro]

I added the file nut.yml, so that generating images in a docker image, on GPU, can be as easy as running the command nut run in the repo. This address issues/PR such as #151

Nut is a tool to run commands in a docker container, while supporting nvidia devices, volumes, etc. Here, nut.yml declares the following settings:

• use docker image matthieudelaro/cuda-torch-plus
• mount the current directory as /opt/style
• mount nvidia GPUs in the container
• run the neural network with 1000 iterations, computing a new image from style_image.jpg and content_image.jpg

Note that Docker supports GPUs on Linux only, and that Nut relies on nvidia-docker-plugin (details).

In case GPUs are not available, nut will display a warning message and run the neural network.

To run the neural network with a custom command, you can use --exec flag. For example:

nut --exec='th neural_style.lua -style_image examples/inputs/picasso_selfport1907.jpg -content_image examples/inputs/brad_pitt.jpg -output_image profile.png -model_file models/nin_imagenet_conv.caffemodel -proto_file models/train_val.prototxt -gpu 0 -backend clnn -num_iterations 1000 -seed 123 -content_layers relu0,relu3,relu7,relu12 -style_layers relu0,relu3,relu7,relu12 -content_weight 10 -style_weight 1000 -image_size 512 -optimizer adam'