import tensorflow as tf
import os
from variant import VARIANT, get_env_from_name, get_policy, get_train, get_eval
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
import ENV.env
if __name__ == '__main__':
root_dir = VARIANT['log_path']
if VARIANT['train']:
for i in range(VARIANT['start_of_trial'],
VARIANT['start_of_trial'] + VARIANT['num_of_trials']):
print(VARIANT)
VARIANT['log_path'] = root_dir + '/' + str(i)
train = get_train(VARIANT['algorithm_name'])
train(VARIANT)
tf.reset_default_graph()
else:
eval = get_eval(VARIANT['algorithm_name'])
eval(VARIANT)
import tensorflow as tf
import os
from variant import VARIANT, get_env_from_name, get_train, get_eval
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
if __name__ == "__main__":
root_dir = VARIANT["log_path"]
if VARIANT["train"]:
for i in range(
VARIANT["start_of_trial"],
VARIANT["start_of_trial"] + VARIANT["num_of_trials"],
):
VARIANT["log_path"] = root_dir + "/" + str(i)
print("logging to " + VARIANT["log_path"])
train = get_train(VARIANT["algorithm_name"])
train(VARIANT)
tf.reset_default_graph()
else:
print("evaluation")
eval = get_eval(VARIANT["algorithm_name"])
eval(VARIANT)