OpenAI Gym website
Gym is a toolkit for developing and comparing reinforcement learning algorithms. It supports teaching agents everything from walking to playing games like Pong or Pinball.
OpenAI Baseline github
OpenAI Baselines is a set of high-quality implementations of reinforcement learning algorithms.
1 安装anaconda&C++编译器
使用Conda的目的是使用虚拟环境,C++编译器可以直接安装Visual Studio 2019。
2 安装Gym
使用 anaconda powershell promopt 来输入:
conda create -n <env-name> python=3.6
创建虚拟环境
conda activate <env-name>
激活环境
(可选)下载git
conda install git
(可选)如果你需要保存训练的图像,还需要下载ffmpeg包:
conda install -c conda-forge ffmpeg
pip install gym
安装gym
或从源码编译:
git clone https://github.com/openai/gym.git cd gym pip install -e .xx
源码下载位置为promot当前路径
编译完成后使用
python examples/agents/random_agent.py
测试,将会出现经典的 cart-pole 的场景
3 安装Baseline
先安装tensorflow
pip install tensorflow==1.14
或pip install tensorflow-gpu==1.14
在baseline GitHub页面下载解压到目标文件夹
或用git命令下载
git clone https://github.com/openai/baselines.git
cd baselines
pip install -e .
运行编译成功后,在baselines目录下运行一个DQN算法跑cartpole的例子:
python baselines/deepq/experiments/train_cartpole.py
若报错
ImportError: Could not find 'cudart64_100.dll'. TensorFlow requires that this DLL be installed in a directory that is named in your %PATH% environment variable. Download and install CUDA 10.0 from this URL: https://developer.nvidia.com/cuda-90-download-archive
可以参考[TensorFlow] windows同时安装cuda10.0和cuda10.1
将会用DQN训练agent 100ksteps, 当训练完成后,你可以运行:
python baselines/deepq/experiments/enjoy_cartpole.py
来观看训练结果