|
|
@@ -7,7 +7,7 @@ designed to be particularly simple, lightweight and fast. The code has very few
|
|
|
dependencies, making it less likely to break or fail to install. It loads no
|
|
|
external sprites/textures, and it can run at up to 5000 FPS on a Core i7
|
|
|
laptop, which means you can run your experiments faster. A known-working RL
|
|
|
-implementation can be found [in this repository](https://github.com/lcswillems/pytorch-a2c-ppo).
|
|
|
+implementation can be found [in this repository](https://github.com/lcswillems/torch-rl).
|
|
|
|
|
|
Requirements:
|
|
|
- Python 3.5+
|
|
|
@@ -56,12 +56,12 @@ The environment being run can be selected with the `--env-name` option, eg:
|
|
|
|
|
|
## Reinforcement Learning
|
|
|
|
|
|
-If you want to train an agent with reinforcement learning, I recommend using the code found in the [pytorch-a2c-ppo](https://github.com/lcswillems/pytorch-a2c-ppo) repository. This code has been tested and is known to work with this environment. The default hyper-parameters are also known to converge.
|
|
|
+If you want to train an agent with reinforcement learning, I recommend using the code found in the [torch-rl](https://github.com/lcswillems/torch-rl) repository. This code has been tested and is known to work with this environment. The default hyper-parameters are also known to converge.
|
|
|
|
|
|
A sample training command is:
|
|
|
|
|
|
```
|
|
|
-cd pytorch-a2c-ppo
|
|
|
+cd torch-rl
|
|
|
python3 -m scripts.train --env MiniGrid-Empty-8x8-v0 --algo ppo
|
|
|
```
|
|
|
|
Lucas Willems