Stable Baselines/用户向导/开始
作者:互联网
大多数强化学习算法包都试图采用sklearn风格语法。
下面是一个简单的案例,展示如何在Cartpole环境中训练和运行PPO2.
import gym
from stable_baselines.common.policies import MlpPolicy
from stable_baselines.common.vec_env import DummyVecEnv
from stable_baselines import PPO2
env = gym.make('CartPole-v1')
env = DummyVecEnv([lambda: env]) # The algorithms require a vectorized environment to run
model = PPO2(MlpPolicy, env, verbose=1)
model.learn(total_timesteps=10000)
obs = env.reset()
for i in range(1000):
action, _states = model.predict(obs)
obs, rewards, dones, info = env.step(action)
env.render()
或者,如果环境已在Gym注册、策略也已注册,仅仅用liner训练一个模型:
# 用一行代码定义并训练一个RL agent
from stable_baselines import PPO2
model = PPO2('MlpPolicy', 'CartPole-v1').learn(10000)
标签:stable,向导,baselines,env,Stable,import,model,PPO2,Baselines 来源: https://blog.csdn.net/The_Time_Runner/article/details/97519520