Much of the day I was trying to figure out the dimensions of convoluted networks. Then I saw that TensorFlow does these things on its own. Neat.
I understand most of the code I wrote for TensorFlow implementation of CNNs and Multilayer perceptron networks. I should get better at graphs as well as data visualisation. It's rather impressive that a simple 50 lines of tensorflow program can identify MNIST handwritten numbers at 99% accuracy.
I also realised I've forgotten my awks and seds as well as shell scripts. Struggle is real.
Mood: Optimistic, curious.
Must stay off the reddit and twitter though.