python error

[buraq1]

Hi, I am working on an algorithm called Augmented random search (ARS) this is its main code:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

def mkdir(base, name):     path = os.path.join(base, name)     if not os.path.exists(path):         os.makedirs(path)         return path   work_dir = mkdir("exp", "brs") monitor_dir = mkdir(work_dir, "monitor")   hp = Hp() np.random.seed(hp.seed) env = gym.make(hp.env_name) env = wrappers.Monitor(env, monitor_dir, force=True) nb_inputs = env.observation_space.shape[0] nb_outputs = env.action_space.shape[0] policy = Policy(nb_inputs, nb_outputs) normalizer = Normalizer(nb_inputs) train(env, policy, normalizer, hp)

and I get this error, Can you help me solve this problem?

Error:
Traceback (most recent call last):
line 134, in <module>
    monitor_dir = mkdir(work_dir, "monitor")
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^
   line 128, in mkdir
    path = os.path.join(base, name)
           ^^^^^^^^^^^^^^^^^^^^^^^^
 line 108, in join
TypeError: expected str, bytes or os.PathLike object, not NoneType

Process finished with exit code 1

Yoriz write Oct-26-2024, 09:41 AM:
Please post all code, output and errors (in it's entirety) between their respective tags. Refer to BBCode help topic on how to post. Use the "Preview Post" button to ensure the code is presented as expected before hitting the "Post Reply/Thread" button.

[Yoriz]

I expect it is because inside mkdir(base, name) a path is only returned when if not os.path.exists(path) is true.
Change the indentation so path is always returned.

1 2 3 4 5

def mkdir(base, name):     path = os.path.join(base, name)     if not os.path.exists(path):         os.makedirs(path)     return path

[buraq1]

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

# AI 2018   # Importing the libraries import os import numpy as np import gym from gym import wrappers import pybullet_envs   # Setting the Hyper Parameters   class Hp():       def __init__(self):         self.nb_steps = 1000         self.episode_length = 1000         self.learning_rate = 0.02         self.nb_directions = 16         self.nb_best_directions = 16         assert self.nb_best_directions <= self.nb_directions         self.noise = 0.03         self.seed = 1         self.env_name = ' '     # Normalizing the states   class Normalizer():       def __init__(self, nb_inputs):         self.n = np.zeros(nb_inputs)         self.mean = np.zeros(nb_inputs)         self.mean_diff = np.zeros(nb_inputs)         self.var = np.zeros(nb_inputs)       def observe(self, x):         self.n += 1.         last_mean = self.mean.copy()         self.mean += (x - self.mean) / self.n         self.mean_diff += (x - last_mean) * (x - self.mean)         self.var = (self.mean_diff / self.n).clip(min=1e-2)       def normalize(self, inputs):         obs_mean = self.mean         obs_std = np.sqrt(self.var)         return (inputs - obs_mean) / obs_std     # Building the AI   class Policy():       def __init__(self, input_size, output_size):         self.theta = np.zeros((output_size, input_size))       def evaluate(self, input, delta=None, direction=None):         if direction is None:             return self.theta.dot(input)         elif direction == "positive":             return (self.theta + hp.noise * delta).dot(input)         else:             return (self.theta - hp.noise * delta).dot(input)       def sample_deltas(self):         return [np.random.randn(*self.theta.shape) for _ in range(hp.nb_directions)]       def update(self, rollouts, sigma_r):         step = np.zeros(self.theta.shape)         for r_pos, r_neg, d in rollouts:             step += (r_pos - r_neg) * d         self.theta += hp.learning_rate / (hp.nb_best_directions * sigma_r) * step     # Exploring the policy on one specific direction and over one episode   def explore(env, normalizer, policy, direction=None, delta=None):     state = env.reset()     done = False     num_plays = 0.     sum_rewards = 0     while not done and num_plays < hp.episode_length:         normalizer.observe(state)         state = normalizer.normalize(state)         action = policy.evaluate(state, delta, direction)         state, reward, done, _ = env.step(action)         reward = max(min(reward, 1), -1)         sum_rewards += reward         num_plays += 1     return sum_rewards     # Training the AI   def train(env, policy, normalizer, hp):     for step in range(hp.nb_steps):           # Initializing the perturbations deltas and the positive/negative rewards         deltas = policy.sample_deltas()         positive_rewards = [0] * hp.nb_directions         negative_rewards = [0] * hp.nb_directions           # Getting the positive rewards in the positive directions         for k in range(hp.nb_directions):             positive_rewards[k] = explore(env, normalizer, policy, direction="positive", delta=deltas[k])           # Getting the negative rewards in the negative/opposite directions         for k in range(hp.nb_directions):             negative_rewards[k] = explore(env, normalizer, policy, direction="negative", delta=deltas[k])           # Gathering all the positive/negative rewards to compute the standard deviation of these rewards         all_rewards = np.array(positive_rewards + negative_rewards)         sigma_r = all_rewards.std()           # Sorting the rollouts by the max(r_pos, r_neg) and selecting the best directions         scores = {k: max(r_pos, r_neg) for k, (r_pos, r_neg) in enumerate(zip(positive_rewards, negative_rewards))}         order = sorted(scores.keys(), key=lambda x: scores[x], reverse=True)[:hp.nb_best_directions]         rollouts = [(positive_rewards[k], negative_rewards[k], deltas[k]) for k in order]           # Updating our policy         policy.update(rollouts, sigma_r)           # Printing the final reward of the policy after the update         reward_evaluation = explore(env, normalizer, policy)         print('Step:', step, 'Reward:', reward_evaluation)   def mkdir(base, name):     path = os.path.join(base, name)     if not os.path.exists(path):         os.makedirs(path)         return path   work_dir = mkdir("exp", "brs") monitor_dir = mkdir(work_dir, "monitor")   hp = Hp() np.random.seed(hp.seed) env = gym.make(hp.env_name) env = wrappers.Monitor(env, monitor_dir, force=True) nb_inputs = env.observation_space.shape[0] nb_outputs = env.action_space.shape[0] policy = Policy(nb_inputs, nb_outputs) normalizer = Normalizer(nb_inputs) train(env, policy, normalizer, hp)

Error:
C:\Users\alnaseem\AppData\Local\Programs\Python\Python312\python.exe "C:\Users\alnaseem\PycharmProjects\ARS\ARS MSA.py" 
Traceback (most recent call last):
  File "C:\Users\alnaseem\PycharmProjects\ARS\ARS MSA.py", line 134, in <module>
    monitor_dir = mkdir(work_dir, "monitor")
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\alnaseem\PycharmProjects\ARS\ARS MSA.py", line 128, in mkdir
    path = os.path.join(base, name)
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "<frozen ntpath>", line 108, in join
TypeError: expected str, bytes or os.PathLike object, not NoneType

Process finished with exit code 1

Output:
No output yet

[bowlofred]

Your function:

1 2 3 4 5

def mkdir(base, name):     path = os.path.join(base, name)     if not os.path.exists(path):         os.makedirs(path)         return path

Note that if the path does not already exist, it tries to make the path and returns that path.
But if the path is already present, the function returns None.

Then later:

1 2	work_dir = mkdir("exp", "brs") monitor_dir = mkdir(work_dir, "monitor")

You use the return value of your mkdir without checking it. It could be None.