Commit 26f62896 by Werner Duvaud

Improve performance

parent fd660a25
...@@ -37,16 +37,16 @@ Testing Lunar Lander : ...@@ -37,16 +37,16 @@ Testing Lunar Lander :
![lunarlander training preview](https://github.com/werner-duvaud/muzero-general/blob/master/docs/lunarlander_training_preview.png) ![lunarlander training preview](https://github.com/werner-duvaud/muzero-general/blob/master/docs/lunarlander_training_preview.png)
## Code structure
![code structure](https://github.com/werner-duvaud/muzero-general/blob/master/docs/how-it-works-werner-duvaud.png)
## Games already implemented with pretrained network available ## Games already implemented with pretrained network available
* Cartpole * Cartpole
* Lunar Lander * Lunar Lander
* Connect4 * Connect4
## Code structure
![code structure](https://github.com/werner-duvaud/muzero-general/blob/master/docs/how-it-works-werner-duvaud.png)
## Getting started ## Getting started
### Installation ### Installation
......
...@@ -67,16 +67,14 @@ class MuZeroConfig: ...@@ -67,16 +67,14 @@ class MuZeroConfig:
Returns: Returns:
Positive float. Positive float.
""" """
# if trained_steps < 0.2 * self.training_steps: if trained_steps < 0.2 * self.training_steps:
# return float('inf') return float('inf')
# if trained_steps < 0.5 * self.training_steps: if trained_steps < 0.5 * self.training_steps:
# return 0.8 return 0.8
# elif trained_steps < 0.75 * self.training_steps: elif trained_steps < 0.75 * self.training_steps:
# return 0.5 return 0.5
# else: else:
# return 0.25 return 0.25
return 1
class Game: class Game:
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment