CLIML - an example (cli) python package for machine learning
It is usual that ML projects start from jupyter notebook, but the path of transforming them into a standalone applications is not clear.
This repo contains an example of a callable machine learning python module that can be executed from the command line.
-
Clone the repo
-
Install the package
pip install . -
Generate some sample data
climl datagen > iris.csv -
Train the model
climl train iris.csv model.pkl -
Generate inference data
climl datagen --only-inference > iris2.csv -
Do inference
climl predict model.pkl iris2.csv
The idea of this packages is to showcase machine learning pipelines using the command line
climl datagen | climl train > model.pkl
Or even more ambitious using file descriptors, wich will allow you to execute everythin in one line
climl predict <(climl datagen | climl train) <(climl datagen --only-inference)
- Include tests
- Add CI/CD
- Generate a Dockerfile