Skip to content

MimicByte/QTable

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 

Repository files navigation

QTable

An interactive Q Table program to help teach reinforcement learning. Current Release Here

Editor Controls

Arrow Keys: Changes map size

PageUp: Cycles tool forward

PageDown: Cycles tool backward

Editor UI

Discount: How much value an item retains over each square. A value closer to 1 means that items barely lose value over distance while a value close to 0 means that items will lose value over distance quickly.

Learning Rate: How much the agent values new information over old information.

Numtrain: The number of times the agent will train before displaying results to the user.

Editor Tools

Items: Items consist of the diamond, emerald, coin, and fire. Items can be given a value with the value painter to add rewards and penalties to the map.

Walls: There are two walls, right and down. Placing a right wall also places a left wall in the square to the right, and placing a down wall also places an up wall in the square below.

Agent: The agent is represented by a rounded purple square. Setting the agent position sets the position where the agent will start when running the game. There can only be one agent at a time.

Win or Loss: The small flag can be placed over any tile to make that tile end the the game when the agent reaches it.

Erasers: Each eraser can erase the map items that are shown on the eraser.

Value Painter: The value painter displays the current values of all items and adds a UI element at the top left to set the value of the cursor. After the cursor value is set, click an item to set its value.

Note: All enclosed spaces in the map must have at least one win or loss state!

About

An interactive Q Table program to help teach reinforcement learning

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors