35 lines
907 B
Markdown
35 lines
907 B
Markdown
|
# tiny-rl
|
||
|
>**_NOTE_**: Currently only stable baselines is supported out of the box
|
||
|
|
||
|
The goal of this repository is to make an easy-to-use framework that wraps the QPD algorithm to compress models of RL policies.
|
||
|
|
||
|
|
||
|
## Installation
|
||
|
First these libraries are required to install.
|
||
|
``` sh
|
||
|
sudo apt update
|
||
|
sudo apt install libpng-dev libjpeg-dev zlib1g-dev
|
||
|
```
|
||
|
Afterwards install the python library.
|
||
|
>**_Important_**: Make sure u use python version 3.10.*, setup tools version 65.0 and wheel 0.38.0
|
||
|
``` sh
|
||
|
conda create -n compression python=3.10
|
||
|
conda activate compression
|
||
|
python -m pip install setuptools==65.0 wheel==0.38.4
|
||
|
python -m pip install -e .
|
||
|
```
|
||
|
|
||
|
## Currently supported
|
||
|
- A2C
|
||
|
- PPO
|
||
|
- DQN
|
||
|
- QRDQN
|
||
|
- SAC
|
||
|
|
||
|
## Cheetah env installation if used with conda
|
||
|
``` sh
|
||
|
conda install -c conda-forge xorg-libx11
|
||
|
conda install -c conda-forge glew
|
||
|
conda install -c conda-forge mesalib
|
||
|
conda install -c menpo glfw3
|
||
|
```
|