import gym
from agents.policy_gradient_agents.PPO import PPO
from agents.actor_critic_agents.DDPG import DDPG
from TD3 import TD3
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
config = Config()
config.seed = 1
config.environment = gym.make("MountainCarContinuous-v0")
config.num_episodes_to_run = 450
config.file_to_save_data_results = None
config.file_to_save_results_graph = None
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 3
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
config.hyperparameters = {
"Policy_Gradient_Agents": {
"learning_rate": 0.05,
"linear_hidden_units": [30, 15],
"final_layer_activation": "TANH",
"learning_iterations_per_round": 10,
"discount_rate": 0.9,
from agents.hierarchical_agents.SNN_HRL import SNN_HRL
from agents.policy_gradient_agents.PPO import PPO
from agents.policy_gradient_agents.REINFORCE import REINFORCE
from environments.FaceDiscreete import FaceEnvironementDiscreete
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
config = Config()
config.seed = 1
config.environment = FaceEnvironementDiscreete(
"../weights/blg_small_12_5e-06_5e-05_2_8_small_big_noisy_first_True_512")
config.num_episodes_to_run = 5
config.file_to_save_data_results = "Data_and_Graphs/FaceDiscreete.pkl"
config.file_to_save_results_graph = "Data_and_Graphs/FaceDiscreete.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 2
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
actor_critic_agent_hyperparameters = {
"Actor": {
"learning_rate": 0.0003,
"linear_hidden_units": [64, 64],
from environments.Long_Corridor_Environment import Long_Corridor_Environment
from models.DQN_agents.DQN import DQN
from models.Trainer import Trainer
from models.hierarchical_agents.SNN_HRL import SNN_HRL
from models.hierarchical_agents.h_DQN import h_DQN
from utilities.data_structures.Config import Config
config = Config()
config.seed = 1
config.env_parameters = {"stochasticity_of_action_right": 0.5}
config.environment = Long_Corridor_Environment(
stochasticity_of_action_right=config.
env_parameters["stochasticity_of_action_right"])
config.num_episodes_to_run = 10000
config.file_to_save_data_results = "Data_and_Graphs/Long_Corridor_Results_Data.pkl"
config.file_to_save_results_graph = "Data_and_Graphs/Long_Corridor_Results_Graph.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 3
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
config.hyperparameters = {
"h_DQN": {
"CONTROLLER": {
"batch_size":
256,
from agents.actor_critic_agents.SAC_Discrete import SAC_Discrete
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
from environments.DMP_Env_1D_dynamic import deep_mobile_printing_1d1r
config = Config()
config.seed = 1
# 0: sin
# 1: gaussian
# 2: step
config.environment = deep_mobile_printing_1d1r()
config.num_episodes_to_run = 10000
config.file_to_save_data_results = "results/data_and_graphs/Cart_Pole_Results_Data.pkl"
config.file_to_save_results_graph = "results/data_and_graphs/Cart_Pole_Results_Graph.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = True
config.overwrite_existing_results_file = True
config.randomise_random_seed = True
config.save_model = False
config.hyperparameters = {
"Actor_Critic_Agents": {
"learning_rate": 0.005,
"linear_hidden_units": [20, 10],
sys.path.append(dirname(dirname(abspath(__file__))))
import gym
from agents.actor_critic_agents.A2C import A2C
from agents.actor_critic_agents.A3C import A3C
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
config = Config()
config.seed = 1
config.environment = gym.make("gym_boxworld:boxworldRandomSmall-v0")
config.num_episodes_to_run = int(1e3)
config.file_to_save_data_results = "results/data_and_graphs/Boxworld_Results_Data.pkl"
config.file_to_save_results_graph = "results/data_and_graphs/Boxworld_Results_Graph.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = True
config.hyperparameters = {
"Actor_Critic_Agents": {
"learning_rate": 0.0001,
# "linear_hidden_units": [20, 10],
from DDQN import DDQN
from HRL import HRL
from SAC_Discrete import SAC_Discrete
from hierarchical_agents.SNN_HRL import SNN_HRL
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
from agents.DQN_agents.DQN import DQN
from agents.hierarchical_agents.h_DQN import h_DQN
config = Config()
config.seed = 1
config.environment = gym.make("Taxi-v2")
config.env_parameters = {}
config.num_episodes_to_run = 1000
config.file_to_save_data_results = "data_and_graphs/Taxi_data.pkl"
config.file_to_save_results_graph = "data_and_graphs/Taxi_graph.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 3
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
linear_hidden_units = [32, 32]
learning_rate = 0.001
buffer_size = 100000
batch_size = 256
config.seed = 1
config.environment = deep_mobile_printing_3d1r(plan_choose=PALN_CHOICE)
config.num_episodes_to_run = 5000
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = True
config.GPU = "cuda:1"
config.overwrite_existing_results_file = True
config.randomise_random_seed = False
config.save_model = False
OUT_FILE_NAME = "SAC_3d_" + PLAN_NAME + "_seed_" + str(config.seed)
config.save_model_path = "/mnt/NAS/home/WenyuHan/SNAC/SAC/3D/dynamic/" + OUT_FILE_NAME + "/"
config.file_to_save_data_results = "/mnt/NAS/home/WenyuHan/SNAC/SAC/3D/dynamic/" + OUT_FILE_NAME + "/" + "Results_Data.pkl"
config.file_to_save_results_graph = "/mnt/NAS/home/WenyuHan/SNAC/SAC/3D/dynamic/" + OUT_FILE_NAME + "/" + "Results_Graph.png"
if os.path.exists(config.save_model_path) == False:
os.makedirs(config.save_model_path)
config.hyperparameters = {
"Actor_Critic_Agents": {
"learning_rate": 0.0005,
"linear_hidden_units": [64, 256, 64],
"final_layer_activation": ["SOFTMAX", None],
"gradient_clipping_norm": 5.0,
"discount_rate": 0.99,
"epsilon_decay_rate_denominator": 1.0,
"normalise_rewards": False,
"exploration_worker_difference": 2.0,
"clip_rewards": True,
if __name__ == '__main__':
from utilities.data_structures.Config import Config
## envs import ##
from environments.carla_enviroments import env_v1_ObstacleAvoidance
# net = q_network_toa(n_action=4)
# net.to('cuda')
# input = torch.rand(size=(10, 3, 224, 224)).to('cuda')
# q1, q2 = net(input)
config = Config()
config.seed = 1
config.environment = gym.make("ObstacleAvoidance-v0")
config.num_episodes_to_run = 2000
config.file_to_save_data_results = "C:/my_project/Deep-Reinforcement-Learning-Algorithms-with-PyTorch/results/data_and_graphs/carla_obstacle_avoidance/data.pkl"
config.file_to_save_results_graph = "C:/my_project/Deep-Reinforcement-Learning-Algorithms-with-PyTorch/results/data_and_graphs/carla_obstacle_avoidance/data.png"
config.show_solution_score = False
config.visualise_individual_results = True
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = True
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = True
config.resume = True
config.resume_path = ''
config.backbone_pretrain = True
from gym.wrappers import FlattenDictWrapper
from agents.DQN_agents.DQN_HER import DQN_HER
from Bit_Flipping_Environment import Bit_Flipping_Environment
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
from agents.DQN_agents.DQN import DQN
config = Config()
config.seed = 1
config.environment = Bit_Flipping_Environment(14)
config.num_episodes_to_run = 4500
config.file_to_save_data_results = None #"Data_and_Graphs/Bit_Flipping_Results_Data.pkl"
config.file_to_save_results_graph = None #"Data_and_Graphs/Bit_Flipping_Results_Graph.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 3
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
config.hyperparameters = {
"DQN_Agents": {
"learning_rate": 0.001,
"batch_size": 128,
"buffer_size": 100000,
"epsilon_decay_rate_denominator": 150,
"discount_rate": 0.999,
"incremental_td_error": 1e-8,
import gym
from agents.DQN_agents.DDQN import DDQN
from agents.hierarchical_agents.HRL.HRL import HRL
from agents.hierarchical_agents.HRL.Model_HRL import Model_HRL
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
config = Config()
config.seed = 1
config.environment = gym.make("CartPole-v0")
config.env_parameters = {}
config.num_episodes_to_run = 500
config.file_to_save_data_results = "data_and_graphs/hrl_experiments/Cart_Pole_data.pkl"
config.file_to_save_results_graph = "data_and_graphs/hrl_experiments/Cart_Poke.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 10
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
config.load_model = False
# Loss is not drawing a random sample! otherwise wouldnt jump around that much!!
linear_hidden_units = [32, 32]
learning_rate = 0.005 # 0.001 taxi
from agents.DQN_agents.DQN_HER import DQN_HER
from environments.j2n6s300.HER_env_tf import j2n6s300_Environment
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
from datetime import datetime
now = datetime.now() # current date and time
num_episodes_to_run = 500
eps_decay_rate_denom = round(num_episodes_to_run/6)
config = Config()
config.seed = 1
config.environment = j2n6s300_Environment()
config.num_episodes_to_run = num_episodes_to_run
config.file_to_save_data_results = "Data_and_Graphs/{}jaco.pkl".format(now.strftime("%Y-%m-%d_%H-%M-%S_"))
config.file_to_save_results_graph = "Data_and_Graphs/{}jaco.png".format(now.strftime("%Y-%m-%d_%H-%M-%S_"))
config.show_solution_score = False
config.visualise_results_while_training = True
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = True
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.load_model = False
config.load_model_path = "Models/.pt"
config.save_model = False
#config.save_model_path = "Models/{}model.pt".format(now.strftime("%Y-%m-%d_%H-%M-%S_"))
config.save_model_path = "Models/DQN_HER_demo_curr.pt"
from agents.DQN_agents.Dueling_DDQN import Dueling_DDQN
from agents.actor_critic_agents.SAC_Discrete import SAC_Discrete
from agents.actor_critic_agents.A3C import A3C
from agents.policy_gradient_agents.PPO import PPO
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
from agents.DQN_agents.DDQN import DDQN
from agents.DQN_agents.DDQN_With_Prioritised_Experience_Replay import DDQN_With_Prioritised_Experience_Replay
from agents.DQN_agents.DQN import DQN
from agents.DQN_agents.DQN_With_Fixed_Q_Targets import DQN_With_Fixed_Q_Targets
config = Config()
config.seed = 1
config.environment = gym.make("CartPole-v0")
config.num_episodes_to_run = 450
config.file_to_save_data_results = "C:\my_project\Deep-Reinforcement-Learning-Algorithms-with-PyTorch\\results/data_and_graphs/Cart_Pole_Results_Data.pkl"
config.file_to_save_results_graph = "C:\my_project\Deep-Reinforcement-Learning-Algorithms-with-PyTorch\\results/data_and_graphs/Cart_Pole_Results_Graph.png"
config.show_solution_score = False
config.visualise_individual_results = True
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = True
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = True
config.hyperparameters = {
"DQN_Agents": {
"learning_rate": 0.01,
config.resume = False
config.resume_path = 'E:\\reinforcement-learning-based-driving-decision-in-Carla\\results\Models\ObstacleAvoidance-v0\DDQN with Prioritised Replay\\20200611150242\\rolling_score_68.0417.model'
config.backbone_pretrain = False
config.force_explore_mode = True
config.force_explore_stare_e = 0.2 ## when the std of rolling score in last 10 window is smaller than this val, start explore mode
config.force_explore_rate = 0.95 ## only when the current score bigger than 0.8*max(rolling score[-10:]), forece expolre
## data and graphs save dir ##
data_results_root = os.path.join(
os.path.dirname(__file__) + "/data_and_graphs/carla_obstacle_avoidance",
config.log_base)
while os.path.exists(data_results_root):
data_results_root += '_'
os.makedirs(data_results_root)
config.file_to_save_data_results = os.path.join(data_results_root, "data.pkl")
config.file_to_save_results_graph = os.path.join(data_results_root, "data.png")
config.hyperparameters = {
"DQN_Agents": {
"learning_rate": 1e-1,
"batch_size": 256,
"buffer_size": 20000,
"epsilon": 1.0,
"epsilon_decay_rate_denominator": 1.,
"discount_rate": 0.9,
"tau": 0.01,
"alpha_prioritised_replay": 0.6,
"beta_prioritised_replay": 0.1,
"incremental_td_error": 1e-8,
"update_every_n_steps": 1,
height = 15
width = 15
random_goal_place = False
num_possible_states = (height * width)**(1 + 1 * random_goal_place)
embedding_dimensions = [[num_possible_states, 20]]
print("Num possible states ", num_possible_states)
config.environment = Four_Rooms_Environment(
height,
width,
stochastic_actions_probability=0.0,
random_start_user_place=True,
random_goal_place=random_goal_place)
config.num_episodes_to_run = 1000
config.file_to_save_data_results = "data_and_graphs/Four_Rooms.pkl"
config.file_to_save_results_graph = "data_and_graphs/Four_Rooms.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
config.hyperparameters = {
"DQN_Agents": {
"linear_hidden_units": [30, 10],
"learning_rate": 0.01,
from environments.isc_environments.SimpleISC import SimpleISC
from utilities.data_structures.Config import Config
from agents.Trainer import Trainer
from agents.actor_critic_agents import A2C, A3C, DDPG, DDPG_HER
from gym.core import Wrapper
from torch.cuda import is_available
config = Config()
config.environment = Wrapper(SimpleISC(mode="DISCRETE"))
config.num_episodes_to_run = 5
config.file_to_save_data_results = "results/data_and_graphs/isc/IllinoisSolarCar_Results_Data.pkl"
config.file_to_save_results_graph = "results/data_and_graphs/isc/IllinoisSolarCar_Results_Graph.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = False
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = is_available()
config.overwrite_existing_results_file = True
config.randomise_random_seed = False
config.save_model = False
config.seed = 0
config.debug_mode = True
config.wandb_log = True
config.hyperparameters = {
from environments.Atari_Environment import make_atari_game
from models.DQN_agents.DDQN import DDQN
from models.Trainer import Trainer
from models.hierarchical_agents.HRL.HRL import HRL
from utilities.data_structures.Config import Config
config = Config()
config.seed = 1
config.environment = make_atari_game("SpaceInvaders-v0")
config.env_parameters = {}
config.num_episodes_to_run = 500
config.file_to_save_data_results = "data_and_graphs/hrl_experiments/Space_Invaders_Data.pkl"
config.file_to_save_results_graph = "data_and_graphs/hrl_experiments/Space_Invaders.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 10
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
# Loss is not drawing a random sample! otherwise wouldnt jump around that much!!
linear_hidden_units = [32, 32]
learning_rate = 0.005 # 0.001 taxi
buffer_size = 1000000
batch_size = 256
batch_norm = False
embedding_dimensionality = 10
from utilities.data_structures.Config import Config
from agents.Trainer import Trainer
from datetime import datetime
import os
now = datetime.now() # current date and time
date_str = now.strftime("%Y-%m-%d_%H-%M-%S")
os.mkdir('Data_and_Graphs/results_' + date_str)
path = 'Data_and_Graphs/results_' + date_str + '/'
config = Config()
config.seed = 1
config.environment = j2n6s300_Environment(proxyID='Env1')
config.num_episodes_to_run = 1
config.file_to_save_config = path + "config.json"
config.file_to_save_data_results = path + "jaco_DDPG-HER.pkl"
config.file_to_save_results_graph = path + "jaco_DDPG-HER.png"
config.show_solution_score = False
config.visualise_results_while_training = True
config.visualise_individual_results = True
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = True
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.load_model = False
config.load_model_path = "Models/model.pt"
config.save_model = True
config.save_model_path = "Models/{}model.pt".format(
now.strftime("%Y-%m-%d_%H-%M-%S_"))
import gym
from agents.actor_critic_agents.DDPG import DDPG
from agents.actor_critic_agents.SAC import SAC
from agents.actor_critic_agents.TD3 import TD3
from agents.hierarchical_agents.DIAYN import DIAYN
from agents.policy_gradient_agents.PPO import PPO
from agents.Trainer import Trainer
from utilities.data_structures.Config import Config
config = Config()
config.seed = 1
config.environment = gym.make("Walker2d-v2")
config.num_episodes_to_run = 400
config.file_to_save_data_results = "data_and_graphs/Walker_Results_Data.pkl"
config.file_to_save_results_graph = "data_and_graphs/Walker_Results_Graph.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 3
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.save_model = False
actor_critic_agent_hyperparameters = {
"Actor": {
"learning_rate": 0.0003,
"linear_hidden_units": [64, 64],
"final_layer_activation": None,
print(symbol)
gym_anytrading.register_new_kzz(symbol)
config.environment = gym.make('kzz-v1')
# config.environment.update_df()
# column_list = ['turn', 'pctChg']
# column_list = ['turn', 'pctChg']
column_list = ["test2"]
# column_list = ["turn", "pctChg", "peTTM", "psTTM", "pcfNcfTTM", "pbMRQ"]
column_list_str = "_".join(column_list)
# config.environment.update_df(fn=None, column_list=column_list)
config.environment.update_df(fn=None, column_list=None)
# config.environment.update_df(fn=lambda df:df.head(100), column_list=column_list)
# config.environment = gym.make("CartPole-v0")
config.num_episodes_to_run = 50
# config.num_episodes_to_run = 450
config.file_to_save_data_results = "results/data_and_graphs/stocks_Results_Data.pkl"
config.file_to_save_results_graph = "results/data_and_graphs/stocks_Results_Graph.png"
config.show_solution_score = False
config.visualise_individual_results = False
config.visualise_overall_agent_results = True
config.standard_deviation_results = 1.0
config.runs_per_agent = 1
config.use_GPU = False
config.overwrite_existing_results_file = False
config.randomise_random_seed = True
config.model_path = r'drive/My Drive/l_gym/Models/%s' % column_list_str
config.save_model = False
config.load_model = True
config.run_test = True
config.run_test_path = r"drive/My Drive/l_gym/data_and_graphs/%s/%s/{}_run_test.png" % (
symbol, column_list_str)
