Yesterday I left an agent running on default settings (20 trials, 0.99 cooling rate) to try and find the ideal agent for playing standard Tetris (parameter 0). Here are my results:
Using best parameter: Mults: {1, 1, 2, 38, 8}, worth 0.53121482649264, over 1091020 steps.
Policy:
Mults: {1, 1, 2, 38, 8}, worth 0.53121482649264, over 1091020 steps.
Mults: {1, [...]

Decided to go into the lab today to tinker about with the agent. I noticed last time that emergency play appears to be either broken or it doesn’t do what I expect it to do, so I’ve turned it off for now and am instead optimising the number of runs for the agent to trial [...]

Finished the mutation tests and code for ParameterSet class. Now all parameters should correctly mutate when told to do so, which still needs to be put in.
Now to tackle the SmartAgent class. I’ve brushed upon it, and most of it isn’t too bad, which is a relief. I thought SubState was more coupled than what [...]

Started on massive refactoring by getting rid of the SubState system completely and all linked code to it.
Created a new class ParameterSet which stores the parameters and the results of the parameter. Also the class can mutate itself to create children.
Laid down the framework for the tests and filled in some of the basic test [...]

Just thinking on my walk over here to the labs about doing a comic of sorts, displaying my thoughts (and fantasies). Like all my ‘projects’ it’ll likely be short lived. But it might be fun. First one: Tripwire or Sliced Bass?