Spaces:
Sleeping
Sleeping
File size: 1,565 Bytes
44db2f9 f05ece6 44db2f9 f05ece6 abff1ef f05ece6 44db2f9 62c6c3b 44db2f9 f05ece6 44db2f9 62c6c3b 44db2f9 |
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 |
import os
import sys
import gym
import matplotlib.pyplot as plt
import torch.multiprocessing as mp
from a3c.discrete_A3C import Net, Worker
from a3c.shared_adam import SharedAdam
from wordle_env.wordle import WordleEnvBase
os.environ["OMP_NUM_THREADS"] = "1"
if __name__ == "__main__":
max_ep = int(sys.argv[1]) if len(sys.argv) > 1 else 100000
env_id = sys.argv[2] if len(sys.argv) > 2 else 'WordleEnv100FullAction-v0'
env = gym.make(env_id)
n_s = env.observation_space.shape[0]
n_a = env.action_space.n
words_list = env.words
word_width = len(env.words[0])
gnet = Net(n_s, n_a, words_list, word_width) # global network
gnet.share_memory() # share the global parameters in multiprocessing
opt = SharedAdam(gnet.parameters(), lr=1e-4, betas=(0.92, 0.999)) # global optimizer
global_ep, global_ep_r, res_queue, win_ep = mp.Value('i', 0), mp.Value('d', 0.), mp.Queue(), mp.Value('i', 0)
# parallel training
workers = [Worker(max_ep, gnet, opt, global_ep, global_ep_r, res_queue, i, env, n_s, n_a, words_list, word_width, win_ep) for i in range(mp.cpu_count())]
[w.start() for w in workers]
res = [] # record episode reward to plot
while True:
r = res_queue.get()
if r is not None:
res.append(r)
else:
break
[w.join() for w in workers]
print("Jugadas:", global_ep.value)
print("Ganadas:", win_ep.value)
plt.plot(res)
plt.ylabel('Moving average ep reward')
plt.xlabel('Step')
plt.show() |