this seems really good. I always like to read code cause it gives a better idea how things works under the hood. when I first wrote a toy neural network I read everything I could on the topic, plus I was taking a ML class where the teacher was a big fan of implementing things from scratch.