MiniTorch is a DIY deep learning library that I developed together with Prof. Sasha Rush in 2020, primarily for the Machine Learning Engineering course at Cornell Tech. MiniTorch is a Python re-implementation of the Torch API, designed to teach machine learning engineers core deep learning concepts such as tensor operations, autograd, and CUDA programming. Full documentaiton is available on the MiniTorch website: https://minitorch.github.io/#
(You may notice that there is a minitorch organization, which has no public members, and a Cornell-Tech-ML organization. The latter is created by us and used by our students in the first offering of the class, while the former one is for the general public. This repo is a public fork of the full student suite available to the general public.)
You can create a fork from the full minitorch student suite. On the MiniTorch website, you can find detailed tutorials and instructions on how to complete it step by step.
We do no release solutions for teaching purpose, but you can use autograders we have built in GitHub Classroom:
- Module 0: https://classroom.github.com/a/qDYKZff9
- Module 1: https://classroom.github.com/a/6TiImUiy
- Module 2: https://classroom.github.com/a/0ZHJeTA0
- Module 3: https://classroom.github.com/a/U5CMJec1
- Module 4: https://classroom.github.com/a/04QA6HZK
- Quizzes: https://classroom.github.com/a/bGcGc12k