HIGHER_DIM_OBS = True
HIGH_INT_DIM = False
if __name__ == "__main__":
logging.basicConfig(level=logging.INFO)
# --- Parse parameters ---
parameters = process_args(sys.argv[1:], Defaults)
if parameters.deterministic:
rng = np.random.RandomState(parameters.seed)
else:
rng = np.random.RandomState()
# --- Instantiate environment ---
env = catcher_env(rng,
higher_dim_obs=HIGHER_DIM_OBS,
reverse=False,
show_game=True)
# --- Instantiate learning algorithm ---
learning_algo = CRAR(env,
rng,
double_Q=True,
high_int_dim=HIGH_INT_DIM,
internal_dim=3)
test_policy = EpsilonGreedyPolicy(learning_algo, env.nActions(), rng,
0.1) #1.)
# --- Instantiate agent ---
agent = NeuralAgent(env,
learning_algo,
HIGH_INT_DIM = False
if __name__ == "__main__":
logging.basicConfig(level=logging.INFO)
# --- Parse parameters ---
parameters = process_args(sys.argv[1:], Defaults)
if parameters.deterministic:
rng = np.random.RandomState(123456)
else:
rng = np.random.RandomState()
# --- Instantiate environment ---
env = catcher_env(rng,
higher_dim_obs=parameters.high_dim_obs,
reverse=False,
show_game=False)
# --- Instantiate learning algorithm ---
learning_algo = CRAR(env,
parameters.rms_decay,
parameters.rms_epsilon,
parameters.momentum,
parameters.clip_norm,
parameters.freeze_interval,
parameters.batch_size,
parameters.update_rule,
rng,
double_Q=True,
high_int_dim=HIGH_INT_DIM,
internal_dim=3)
HIGHER_DIM_OBS = True
HIGH_INT_DIM = False
if __name__ == "__main__":
logging.basicConfig(level=logging.INFO)
# --- Parse parameters ---
parameters = process_args(sys.argv[1:], Defaults)
if parameters.deterministic:
rng = np.random.RandomState(123456)
else:
rng = np.random.RandomState()
# --- Instantiate environment ---
env = catcher_env(rng, higher_dim_obs=HIGHER_DIM_OBS, reverse=False)
# --- Instantiate learning algorithm ---
learning_algo = CRAR(env,
parameters.rms_decay,
parameters.rms_epsilon,
parameters.momentum,
parameters.clip_norm,
parameters.freeze_interval,
parameters.batch_size,
parameters.update_rule,
rng,
double_Q=True,
high_int_dim=HIGH_INT_DIM,
internal_dim=3)