README
¶
Go Coach
A library for training reinforcement learning agents. Its philosophy is to separate the RL machinery from the end user, so that users provide an environment and get back a trained agent who can then act intelligently in that environment, whatever it may be.
It is inspired by (and may at some point wrap) https://github.com/aunum/gold, but with the goal of separating out the "internals" of the training, so it is as easy as possible to apply RL to new situations.
Usage
Run:
go run examples/main.go
To create your own environment, see pkg/coach/environment.go
Directories
¶
| Path | Synopsis |
|---|---|
|
deeplizard
A simple game where a lizard walks around a board and tries to find crickets to eat: https://www.youtube.com/watch?v=qhRNvCVVJaA
|
A simple game where a lizard walks around a board and tries to find crickets to eat: https://www.youtube.com/watch?v=qhRNvCVVJaA |
|
frozenlake
https://gym.openai.com/envs/FrozenLake-v0/
|
https://gym.openai.com/envs/FrozenLake-v0/ |
|
konamicode
Simple game whose only goal is to type the Konami Code (https://en.wikipedia.org/wiki/Konami_Code)
|
Simple game whose only goal is to type the Konami Code (https://en.wikipedia.org/wiki/Konami_Code) |
|
wordle
https://www.powerlanguage.co.uk/wordle/
|
https://www.powerlanguage.co.uk/wordle/ |
|
pkg
|
|
Click to show internal directories.
Click to hide internal directories.