├── README.md Add some info about your project here
├── data Your data files (ignored in version control) and datasets specifications lives here
│ ├── processed Do not ever modify your raw data, do your preprocessing and store the final files here
│ ├── cache Downloaded raw data files lives here
│ └── raw Do not download any files here
│ ├── dataset1
│ │ └── .toml Specifications for downloading data
│ ├── dataset2
│ │ └── .csv Actual dataset with links to binaries like images, audio files, etc... Use git-lfs if this file gets big enough
│ └── dataset3
│ └── .json Actual dataset with links to binaries like images, audio files, etc... Use git-lfs if this file gets big enough
├── logs Your model training logs live here, so model training could be monitored by TensorBoard
├── notebooks Your notebooks live here and their name indicate the order they should be ran. i.e. 01-explore-... and 02-clean-...
├── src Your python files live here
│ ├── __init__.py
│ └── utils
│ ├── __init__.py
│ └── utils.py
├── serving Your model serving classes lives here
│ └── example_model
│ ├── requirements.txt
│ └── user_model.py
├── scripts Your shell scripts that perform various tasks live here
│ └── cleanup.sh Clear your cached data, logs, etc...
└── weights gitignore this directory if you do not want to push your models to git
Conda or pipenv come pre-installed for easy quick use. We recommend using conda though.
First you need to change your current directory to the one where you want your environment config to be tracked.
If you want to use pipenv, initalize your environment by calling pipenv lock
and if you want to use conda, run conda create --name ENV_NAME python=3.7 -y
.
If you are using conda:
conda activate ENV_NAME
ipython kernel install --user --name=ENV_NAME
If you are using pipenv
, virtualenv
, etc... install ipykernel
in your new environment pip install ipykernel
. And then python -m ipykernel install --user --name=ENV_NAME
.
First to find all the available kernel specs jupyter kernelspec list
and then you can run jupyter kernelspec remove KERNAL_NAME
First you need to install kaggle cli pip install kaggle --upgrade
and then generate an API key by going to https://www.kaggle.com//account and select 'Create API Token'
export KAGGLE_USERNAME=username
export KAGGLE_KEY=xxxxxxxxx