In [6]:

    

from timeit import default_timer as timer

from tqdm import tqdm
import numpy as np

from env import Env2048


    

In [7]:

    

def experiment():
    env = Env2048()
    env.reset()
    s = timer()
    n = 1000
    for t in range(n):
        action = env.action_space.sample()
        _, _, done, _ = env.step(action)
        if done:
            env.reset()
    e = timer()
    env.close()
    return n / (e - s)


    

In [8]:

    

apss = []
for i in tqdm(range(20), unit="run"):
    aps = experiment()
    apss.append(aps)
print(f"Mean actions per second : {np.mean(apss)}")


    

    




100%|██████████| 20/20 [00:40<00:00,  2.02s/run]





    




Mean actions per second : 494.65210894465343






    

In [ ]: