How do I get my simple neural network to output values closer to the labels?

Question

I\'m pretty new to reinforcement learning and deep Q learning. I\'m trying to learn by making a tic tac toe DQN as a starter project. I have some experience with Q learning,