powermundsen / ml2048 Goto Github PK

We should only apply the reward and dicount to the max q value of this state

On exiting (closing the game window), the following error appears:

TclError Traceback (most recent call last)
in ()
4
5 our_machine=machine.Machine()
----> 6 our_machine.run()

C:\Users\gurce\Documents\Repos\ml2048\src\machine.py in run(self)
16 self.game.key_down(self.act()) #COMMENT TO RUN EXAMPLE
17 self.game.update_idletasks
---> 18 self.game.update()
19 def act(self):
20 return ("'z'") #DOES NOTHING

C:\Users\gurce\Anaconda3\lib\tkinter_init_.py in update(self)
1169 def update(self):
1170 """Enter event loop until all pending events have been processed by Tcl."""
-> 1171 self.tk.call('update')
1172 def update_idletasks(self):
1173 """Enter event loop until all idle callbacks have been called. This

TclError: can't invoke "update" command: application has been destroyed

Logging slows down the program.

I think it's because it's opening and closing the log file on each iteration.
We should change it so that, for example, it only writes to the log once each episode.
I'm writing this here so that we don't forget.

I'll try to do this myself when I have the time but feel free to fix it if you wan't.