Playing around w/ tinygrad framework. This repo uses the docker environment from CudaConda3 for NVidia GPU support.
To start up the containerized environment, run the following commands to build the tinygrad environment and then start a container w/ a Jupyter notebook environment.
docker build -t tinygrad .
docker run --name tinygrad --rm -p 8888:8888 -d --gpus all tinygrad
If you want to mount your local directory to save your work to disk between docker session, use the volume flag:
docker run -v .:/root --name tinygrad --rm -p 8888:8888 -d --gpus all tinygrad
Then go to http://localhost:8888/jupyter/lab to see you notebook session.