How to operate the neural network: Set variables learnrate and div to 1 for fast, unprecise learning. If gravity goes too low (example: -4000), press M to reset it, which assists the neural network in its learning as it allows it to discover more positions to save into its training data. You can also reset gravity before reaching a large integer in the thousands, as it is beneficial to this network's operation. Another way to assist the network is to set the value of the variable div to 100 and keep learnrate at 1, which was the standard operation for the earlier versions of the network and software. It is good if you want your neural network to learn slowly but precisely. You can press Q to make the system to continuously upload 10,000 mistakes/positions to the network, regardless of whether the plane is touching the ground. And if you want the most precise, slow learning, drag the plane to a suitable position and press SPACE. This makes the plane fly across the screen very slowly, saving every position, moving from left to right. Gravity does not apply, but we do recommend setting div to 100 and learnrate to 1. As well as being able to drag the airplane, you can rotate the aircraft by using the left and right arrow keys to assist its learning. When you press P (this feature is not for performance and just for immersive sound experience), the adaptive sound system activates, causing the plane to sound more realistic. As well as being able to adjust sound, the keys 1 and 2 can be used to adjust the volume of sounds made by the aircraft. Also, if you see a purple square appear and an alert play, it means the neural network is using its training data and levelling out. The variable dat.us counts how many times this has happened. Finally, you can press W to wipe all training data from the network. Kindest regards on using the software, Jacob Harvey, 11, Scratch Programmer.
Person who made plane and me for AI
