Neural Nets from scratch in C++/Cuda

Compile and Run

Using NVCC [Cuda]

❯ nvcc -std=c++17 test.cu -o testfile          
❯ ./testfile              

Using G++ [C++]

❯ g++ -std=c++17 test.cpp -o testfile          
❯ ./testfile              

Checkpoints

Getting started

• Basic Overview of Build/Compile and Make methods in C++/Cuda
• File handling in C++ and writing custom I/O functions
• Add makefile
• Structuring C++ projects
• Object oriented design of neural nets

Feature implementation

• Read_csv implementation in C++
• Dense Layer Implementation
• Activation Function Implementation
• Write Train Function

Misc

• Add Iris Dataset
• Use C++17 as default compilation mode
• Add comments in csv.h