111 lines
4.3 KiB
Python
111 lines
4.3 KiB
Python
#!/usr/bin/env python
|
||
# coding=utf-8
|
||
'''
|
||
Author: John
|
||
Email: johnjim0816@gmail.com
|
||
Date: 2020-09-11 23:03:00
|
||
LastEditor: John
|
||
LastEditTime: 2021-04-29 17:01:08
|
||
Discription:
|
||
Environment:
|
||
'''
|
||
import sys,os
|
||
curr_path = os.path.dirname(__file__)
|
||
parent_path=os.path.dirname(curr_path)
|
||
sys.path.append(parent_path) # add current terminal path to sys.path
|
||
|
||
import gym
|
||
import datetime
|
||
|
||
from envs.gridworld_env import CliffWalkingWapper
|
||
from QLearning.agent import QLearning
|
||
from common.plot import plot_rewards
|
||
from common.utils import save_results,make_dir
|
||
curr_time = datetime.datetime.now().strftime("%Y%m%d-%H%M%S") # obtain current time
|
||
|
||
class QlearningConfig:
|
||
'''训练相关参数'''
|
||
def __init__(self):
|
||
self.algo = 'Qlearning'
|
||
self.env = 'CliffWalking-v0' # 0 up, 1 right, 2 down, 3 left
|
||
self.result_path = curr_path+"/outputs/" +self.env+'/'+curr_time+'/results/' # path to save results
|
||
self.model_path = curr_path+"/outputs/" +self.env+'/'+curr_time+'/models/' # path to save models
|
||
self.train_eps = 300 # 训练的episode数目
|
||
self.eval_eps = 30
|
||
self.gamma = 0.9 # reward的衰减率
|
||
self.epsilon_start = 0.95 # e-greedy策略中初始epsilon
|
||
self.epsilon_end = 0.01 # e-greedy策略中的终止epsilon
|
||
self.epsilon_decay = 200 # e-greedy策略中epsilon的衰减率
|
||
self.lr = 0.1 # learning rate
|
||
|
||
def env_agent_config(cfg,seed=1):
|
||
env = gym.make(cfg.env)
|
||
env = CliffWalkingWapper(env)
|
||
env.seed(seed)
|
||
state_dim = env.observation_space.n
|
||
action_dim = env.action_space.n
|
||
agent = QLearning(state_dim,action_dim,cfg)
|
||
return env,agent
|
||
|
||
def train(cfg,env,agent):
|
||
rewards = []
|
||
ma_rewards = [] # moving average reward
|
||
for i_ep in range(cfg.train_eps):
|
||
ep_reward = 0 # 记录每个episode的reward
|
||
state = env.reset() # 重置环境, 重新开一局（即开始新的一个episode）
|
||
while True:
|
||
action = agent.choose_action(state) # 根据算法选择一个动作
|
||
next_state, reward, done, _ = env.step(action) # 与环境进行一次动作交互
|
||
agent.update(state, action, reward, next_state, done) # Q-learning算法更新
|
||
state = next_state # 存储上一个观察值
|
||
ep_reward += reward
|
||
if done:
|
||
break
|
||
rewards.append(ep_reward)
|
||
if ma_rewards:
|
||
ma_rewards.append(ma_rewards[-1]*0.9+ep_reward*0.1)
|
||
else:
|
||
ma_rewards.append(ep_reward)
|
||
print("Episode:{}/{}: reward:{:.1f}".format(i_ep+1, cfg.train_eps,ep_reward))
|
||
return rewards,ma_rewards
|
||
|
||
def eval(cfg,env,agent):
|
||
# env = gym.make("FrozenLake-v0", is_slippery=False) # 0 left, 1 down, 2 right, 3 up
|
||
# env = FrozenLakeWapper(env)
|
||
rewards = [] # 记录所有episode的reward
|
||
ma_rewards = [] # 滑动平均的reward
|
||
for i_ep in range(cfg.eval_eps):
|
||
ep_reward = 0 # 记录每个episode的reward
|
||
state = env.reset() # 重置环境, 重新开一局（即开始新的一个episode）
|
||
while True:
|
||
action = agent.predict(state) # 根据算法选择一个动作
|
||
next_state, reward, done, _ = env.step(action) # 与环境进行一个交互
|
||
state = next_state # 存储上一个观察值
|
||
ep_reward += reward
|
||
if done:
|
||
break
|
||
rewards.append(ep_reward)
|
||
if ma_rewards:
|
||
ma_rewards.append(ma_rewards[-1]*0.9+ep_reward*0.1)
|
||
else:
|
||
ma_rewards.append(ep_reward)
|
||
print(f"Episode:{i_ep+1}/{cfg.eval_eps}, reward:{ep_reward:.1f}")
|
||
return rewards,ma_rewards
|
||
|
||
if __name__ == "__main__":
|
||
cfg = QlearningConfig()
|
||
env,agent = env_agent_config(cfg,seed=1)
|
||
rewards,ma_rewards = train(cfg,env,agent)
|
||
make_dir(cfg.result_path,cfg.model_path)
|
||
agent.save(path=cfg.model_path)
|
||
save_results(rewards,ma_rewards,tag='train',path=cfg.result_path)
|
||
plot_rewards(rewards,ma_rewards,tag="train",env=cfg.env,algo = cfg.algo,path=cfg.result_path)
|
||
|
||
env,agent = env_agent_config(cfg,seed=10)
|
||
agent.load(path=cfg.model_path)
|
||
rewards,ma_rewards = eval(cfg,env,agent)
|
||
save_results(rewards,ma_rewards,tag='eval',path=cfg.result_path)
|
||
plot_rewards(rewards,ma_rewards,tag="eval",env=cfg.env,algo = cfg.algo,path=cfg.result_path)
|
||
|
||
|