Number generators in most programming languages are not truly random. I trained a neural network to try and find patterns in the number generation and predict what the next number will be. If the success rate of finding a number from 1 till 100 is higher that 1% it means that it could have found a patern
No followers yet
Once you ship this you can't edit the description of the project, but you'll be able to add more devlogs and re-ship it as you add new features!
Added a simple transformer model with RoPE (Rotary Position Embedding). It did not perform any better than the simple neural network with RoPE. Performed with approximately the same success rate as the baseline ~1%.
Added rotational position encodings with huge improvement in guessing. Added graphs and readme
Set everything up, currently working pretty well with ~1.07% success rate as opposed to baseline 1%