Chestrays, like xrays but for chests
We use conda to setup dependencies:
conda env create -f environment.yml
To use the shared kaggle account run the following command in the directory of this README:
export KAGGLE_CONFIG_DIR=`pwd`/secrets
To check if worked run kaggle kernels list -m. It should list some kernels
and not print any warnings or errors.
There are currently two tools in this package: krun and kfetch.
krun takes a notebook (with existing metadata file) and runs it as a kaggle
kernel. Optionally waiting for it to complete and downloading the output.
kfetch takes a notebook (with existing metadata file) and waits for it to
complete running, then fetches the output.
To initially fetch a notebook from a kernel and have the metadata created use:
kaggle kernels pull -p kernels/create-dataturks-nih chestrays/create-pneumonia-dataturks-dataset
This will fetch the kernel called chestrays/create-pneumonia-dataturks-dataset
and put it in the kernels/create-dataturks-nih subdirectory.
To get access to the secrets stored in secrets/ install git-crypt and contact
one of the people who have the key:
- Tim Head