Building Pytorch from scratch

As a part of my deep learning systems course at Carnegie Mellon University, I built a PyTorch-like deep learning framework, called NEEDLE, using only numpy. 

The library includes automatic differentiation, optimization, data loaders, loss functions, and modular layers.

Please refer to this example notebook for implementation details and usage of the needle library. 

<Github repo>

Building ChatGPT from scratch

<Github repo>