-
Notifications
You must be signed in to change notification settings - Fork 0
/
hockey-trainERE.py
146 lines (118 loc) · 6 KB
/
hockey-trainERE.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
import numpy as np
import laserhockey.hockey_env as h_env
import gym
from importlib import reload
import time
import argparse
import datetime
import gym
import itertools
import torch
from sac.sac_better import SAC
from torch.utils.tensorboard import SummaryWriter
from sac.ere_prio_replay import PrioritizedReplay
from sac.replay_memory import ReplayMemory
import copy
parser = argparse.ArgumentParser(description='Soft Actor-Critic Args')
parser.add_argument('--env-name', default="Hockey")
parser.add_argument('--policy', default="Gaussian")
parser.add_argument('--gamma', type=float, default=0.95, metavar='G')
parser.add_argument('--tau', type=float, default=0.005, metavar='G')
parser.add_argument('--lr', type=float, default=0.0003, metavar='G')
parser.add_argument('--alpha', type=float, default=0.2, metavar='G')
parser.add_argument('--automatic_entropy_tuning', type=bool, default=True, metavar='G')
parser.add_argument('--seed', type=int, default=111111, metavar='N')
parser.add_argument('--batch_size', type=int, default=4, metavar='N')
parser.add_argument('--num_steps', type=int, default=1000001, metavar='N')
parser.add_argument('--hidden_size', type=int, default=512, metavar='N')
parser.add_argument('--updates_per_step', type=int, default=1, metavar='N')
parser.add_argument('--start_steps', type=int, default=10000, metavar='N')
parser.add_argument('--target_update_interval', type=int, default=1, metavar='N')
parser.add_argument('--replay_size', type=int, default=1000000, metavar='N')
args = parser.parse_args()
args.cuda =True if torch.cuda.is_available() else False
env = h_env.HockeyEnv(mode=h_env.HockeyEnv.TRAIN_SHOOTING)
# Agent
agent = SAC(env.observation_space.shape[0], env.action_space, args)
time_ = datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S')
#Tesnorboard
writer = SummaryWriter(f"selfstrongplay-runs/ere{time_}_batch_size-{args.batch_size}_gamma-{args.gamma}_tau-{args.tau}_lr-{args.lr}_alpha-{args.alpha}_tuning-{args.automatic_entropy_tuning}_hidden_size-{args.hidden_size}_updatesStep-{args.updates_per_step}_startSteps-{args.start_steps}_targetIntervall-{args.target_update_interval}_replaysize-{args.replay_size}")
# Memory
memory = PrioritizedReplay(args.replay_size)
# memory = ReplayMemory(args.replay_size,args.seed)
# Training Loop
total_numsteps = 0
updates = 0
o = env.reset()
# _ = env.render()
eta_0 = 0.996
eta_T = 1.0
max_ep_len = 1000
c_k_min = 2500 # original = 5000
for i_episode in itertools.count(1):
episode_reward = 0
episode_steps = 0
done = False
state = env.reset()
while not done:
# state = env.obs_agent_two()
if args.start_steps > total_numsteps:
action = env.action_space.sample() # Sample random action
else:
action = agent.select_action(state) # Sample action from policy
a2 = [10,0.,0,0]
# obs_agent2 = env.obs_agent_two()
# a2 = opponent.select_action(obs_agent2, evaluate=True)
next_state, reward, done, _ = env.step(np.hstack([action[0:4],a2[0:4]]))
# env.render()
episode_steps += 1
total_numsteps += 1
episode_reward += reward
# Ignore the "done" signal if it comes from hitting the time horizon.
mask = 1 if episode_steps == 251 else float(not done)
# mask = float(not done)
memory.push(state, action, reward, next_state, mask)
eta_t = eta_0 + (eta_T - eta_0)*(total_numsteps/args.num_steps)
state = next_state
if total_numsteps > args.num_steps:
break
for k in range(1,episode_steps):
c_k = max(int(len(memory)*eta_t**(k*((env.max_timesteps+1)/episode_steps))), c_k_min)
for i in range(args.updates_per_step):
# Update parameters of all the networks
critic_1_loss, critic_2_loss, policy_loss, ent_loss, alpha, memory_ = agent.update_parameters(memory, args.batch_size, updates,c_k=c_k)
memory=memory_
writer.add_scalar('loss/critic_1', critic_1_loss, updates)
writer.add_scalar('loss/critic_2', critic_2_loss, updates)
writer.add_scalar('loss/policy', policy_loss, updates)
writer.add_scalar('loss/entropy_loss', ent_loss, updates)
writer.add_scalar('entropy_temprature/alpha', alpha, updates)
updates += 1
writer.add_scalar('reward/train', episode_reward, i_episode)
print("Episode: {}, total numsteps: {}, episode steps: {}, reward: {}".format(i_episode, total_numsteps, episode_steps, round(episode_reward, 2)))
if i_episode % 10 == 0:
avg_reward = 0.
episodes = 5
for _ in range(episodes):
state = env.reset()
episode_reward = 0
done = False
while not done:
action = agent.select_action(state, evaluate=True)
# obs_agent2 = env.obs_agent_two()
# a2 = opponent.select_action(obs_agent2, evaluate=True)
a2 = [10,0.,0,0]
next_state, reward, done, _ = env.step(np.hstack([action[0:4],a2[0:4]]))
# env.render()
episode_reward += reward
state = next_state
avg_reward += episode_reward
avg_reward /= episodes
if i_episode%100==0:
time_ = datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S')
agent.save_model( "selfstrongplay_models_ere", "hockey", suffix=f"reward-{avg_reward}_episode-"+str(i_episode)+f"_batch_size-{args.batch_size}_gamma-{args.gamma}_tau-{args.tau}_lr-{args.lr}_alpha-{args.alpha}_tuning-{args.automatic_entropy_tuning}_hidden_size-{args.hidden_size}_updatesStep-{args.updates_per_step}_startSteps-{args.start_steps}_targetIntervall-{args.target_update_interval}_replaysize-{args.replay_size}_t-{time_}")
writer.add_scalar('avg_reward/test', avg_reward, i_episode)
print("----------------------------------------")
print("Test Episodes: {}, Avg. Reward: {}".format(episodes, round(avg_reward, 2)))
print("----------------------------------------")
env.close()