nnUNet benchmark

This is a simple orchestration around the nnUNet DL Example

Build the container:

sh build.sh

This example requires a significant data download. So, we're not putting it in a container. We need to pass through a local directory and then run the data download and pre-processing:

# the parameter is where the data will be downloaded to
sh get_data.sh /home/nvidia/data

Now, run the benchmarks - note that there are two; for 2D and 3D data -- ideally you run both:

sh run.sh data_dir n_gpus batch n_dims [--amp]
# on a 1-GPU 16 GB system:
# sh run.sh /home/nvidia/data 1 128 2
# sh run.sh /home/nvidia/data 1 128 2 --amp
# sh run.sh /home/nvidia/data 1 4 3 --amp