Welcome to Pupil’s documentation!!


Pupil

Read the docs PyPI version License Code style: black

What is it?

Pupil [ pyoo-puhl ]: noun

  • a person, usually young, who is learning under the close supervision of a teacher at school, a private tutor, or the like; student.

pupil

  • Python package for labeing data efficiently


:warning: This library is under heavy development.

Have you ever tried to start a cool new machine learning project idea but had to drop it because couldn’t find annotated data? Or realized out it can get expensive really fast to outsource the labeling process? or even tried to do the labeling by yourself but the way you were picking those data points were not efficient and had to label lots of data? Well, I’ve tried them all! It was not fun, and that’s why I made pupil.

Pupil can be used by data scientists/ machine learning engineers who know how to label their data and want to start something quick for proof of concept.

Pupil allows you to do have a database for your embeddings and metadata in one place, and do fast vector similarity searchs. It also has some active learning methods to help you prioritize which data points are more important( Cold and Warm problems), and in the end, you even can labele your data inside the jupyter-lab.

Warning

This library in under heavy development.

Please check the Usage.

ref to Instalation.

Indices and tables