• 0 Posts
  • 10 Comments
Joined 1 year ago
cake
Cake day: June 29th, 2023

help-circle

  • The reason they are blackboxes is because they are function approximators with billions of parameters. Theory has not caught up with practical results. This is why you tune hyperparameters (learning rate, number of layers, number of neurons ina layer, etc.) and have multiple iterations of training to get an approximation of the distribution of the inputs. Training is also sensitive to the order of inputs to the network. A network trained on the same training set but in a different order might converge to an entirely different function. This is why you train on the same inputs in random order over multiple episodes to hopefully average out such variations. They are blackboxes simply because you can’t yet prove theoretically the function it has approximated or converged to given the input.







  • I go back to World of Tanks every once in a while but it is impossible to enjoy it even though I want to. The concept is great, there are tactics involved, but the skill ceiling is just too damn high. So now whenever I get the urge to play, I instead watch a couple of streamers. Skill for his entertaining streams and well, skill (but he gets annoying when he starts raging about not being able to achieve his personal goals), and Dez for the generally chill vibes.

    World of Tanks would have been a great game had it not been plagued with the entire pay to win aspect. And don’t get me started on the whole crew training and retraining nonsense. I hate that aspect of the game, and still don’t understand it. Do I keep the trained crew in a tank I enjoy playing or move them onto the next tier tank and then have terrible stats on the new crew and lose interest in the tank I started enjoying? Uff!