Stepan-Makarenko/DL-from-scratch

Me trying to improve knowleadge of cpp as well as low-level dl concepts

Base description

This is the OOP C++ DL labriary to train and understand Neural Networks. This work is inspired by Andrey Karpathy's awesome repository - https://github.com/karpathy/llm.c

Plan

Matrix2d class and operations on it - ✅

Linear Layer with gradient flow - ✅

Cross Entropy Loss, Sigmoid - ✅

Train 2 layers binary classification model - ✅

Matrix3d class and operations - ✅

Self-attention - 🔲

Layer Norms - 🔲

Gpt2 local inferene - 🔲