os.environ['TF_CPP_MIN_LOG_LEVEL'] = '1'
import tensorflow as tf
from tensorflow.python.keras import Sequential
from tensorflow.python.keras.layers import Dense
from tensorflow.python.keras.activations import softmax, relu, linear, tanh
from modern_keras import reinforce
from line_world import LineWorldEnv, A, num_states
import numpy as np
if __name__ == "__main__":
# tf.compat.v1.disable_eager_execution()
np.set_printoptions(precision=3, suppress=True)
env = LineWorldEnv()
actor_model = Sequential()
actor_model.add(Dense(64, activation=relu))
actor_model.add(Dense(len(A), activation=softmax))
base_line_model = Sequential()
base_line_model.add(Dense(64, activation=tanh))
base_line_model.add(Dense(1, activation=linear))
reinforce(
actor_model,
base_line_model,
env.deep_reset,
env.deep_get_state,
env.deep_is_terminal,
os.environ['TF_DISABLE_MKL'] = '1'
os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '1'
from algorithms import reinforce_with_mask_without_baseline
from brains import SimpleMLPPiSWithMaskBrainReinforce
from line_world import S, A, LineWorldEnv, num_states
import numpy as np
if __name__ == "__main__":
import tensorflow as tf
tf.compat.v1.disable_eager_execution()
brain = SimpleMLPPiSWithMaskBrainReinforce(len(S), len(A), lr=0.0001)
env = LineWorldEnv()
reinforce_with_mask_without_baseline(
env.deep_reset,
env.deep_get_state,
env.deep_get_action_mask,
env.deep_is_terminal,
env.deep_step_with_mask,
brain,
episodes_count=5000,
)
states = env.to_categorical(np.arange(num_states))
mask = env.deep_get_action_mask()
for s_idx, s in enumerate(states):