L2: A deep learning library implemented from scratch in C++

I made this over the summer as a way to learn more about C++ and lower-level machine learning concepts. I wanted to see how much I could build myself, without relying on other libraries to handle things like creating multidimensional arrays. Version 1 of this library focuses only supports a cpu backend at the moment since I'm not familiar enough with c++ to start working with CUDA and cudnn. It also primarily uses pass-by-value to reduce complexity at the result of reduced efficiency. Version 2 of the library will focus on making the Tensor class more efficient. Currently, only the Linear and Sigmoid layers, the MSE loss, and the SGD optimizer have been implemented, but I will add more layers and modules in V2. The library contains a multidimensional array class, Tensor, with support for strided arrays, numpy-style array slicing, broadcasting, and most major math operations (including matrix multiplication!). On top of this, the Parameter, Layer, Sequential, Loss, Optimizer, and Trainer classes allow for running high-level machine learning experiments without worrying about the low-level implementations.