Machine Learning with fast.ai

Setup

I am running most of this locally on a workstation with an nVidia GPU (currently a GTX1080). We can use nvidia-docker to keep a clean environment but we will still need to ensure we have the proper setup on the host machine. These instructions are great to setup the environment and are required to get running before nvidia-docker. See host-setup.sh

Running

Ensure nvidia-docker is setup and working. See host-setup.sh to install it.

GPU Test/Information:

docker run \
    --runtime=nvidia \
    --rm \
    nvidia/cuda nvidia-smi

Run our notebook service:

docker build -t ml-fast-ai .
docker run \
    --runtime=nvidia \
    --rm \
    --name ml-fast-ai \
    --network="host" \
    -v $(pwd)/course-content:/data \
    -p 8888:8888 \
    ml-fast-ai

Monitor GPU Usage:

docker run \
    -ti \
    --runtime=nvidia \
    --rm \
    nvidia/cuda:8.0-cudnn7-runtime-ubuntu16.04 \
    /bin/bash -c "watch -n 1 nvidia-smi"

Practical Deep Learning for Coders - Part 1