Camera-ready cleanup

DDPG.py

@@ -94,8 +94,8 @@ def train(self, replay_buffer, iterations, batch_size=64, discount=0.99, tau=0.0
 
 			# Q target = reward + discount * Q(next_state, pi(next_state))
 			target_Q = self.critic_target(next_state, self.actor_target(next_state))
-			target_Q.volatile = False 
 			target_Q = reward + (done * discount * target_Q)
+			target_Q.volatile = False
 
 			# Get current Q estimate
 			current_Q = self.critic(state, action)
@@ -120,7 +120,7 @@ def train(self, replay_buffer, iterations, batch_size=64, discount=0.99, tau=0.0
 			for param, target_param in zip(self.critic.parameters(), self.critic_target.parameters()):
 				target_param.data.copy_(tau * param.data + (1 - tau) * target_param.data)
 
-			for param, target_param,  in zip(self.actor.parameters(), self.actor_target.parameters()):
+			for param, target_param in zip(self.actor.parameters(), self.actor_target.parameters()):
 				target_param.data.copy_(tau * param.data + (1 - tau) * target_param.data)
 
 

OurDDPG.py

@@ -0,0 +1,133 @@
+import numpy as np
+import torch
+import torch.nn as nn
+from torch.autograd import Variable
+import torch.nn.functional as F
+
+import utils
+
+
+# Re-tuned version of Deep Deterministic Policy Gradients (DDPG)
+# Paper: https://arxiv.org/abs/1509.02971
+
+
+def var(tensor, volatile=False):
+	if torch.cuda.is_available():
+		return Variable(tensor, volatile=volatile).cuda()
+	else:
+		return Variable(tensor, volatile=volatile)
+
+
+class Actor(nn.Module):
+	def __init__(self, state_dim, action_dim, max_action):
+		super(Actor, self).__init__()
+
+		self.l1 = nn.Linear(state_dim, 400)
+		self.l2 = nn.Linear(400, 300)
+		self.l3 = nn.Linear(300, action_dim)
+
+		self.max_action = max_action
+
+
+	def forward(self, x):
+		x = F.relu(self.l1(x))
+		x = F.relu(self.l2(x))
+		x = self.max_action * F.tanh(self.l3(x)) 
+		return x 
+
+
+class Critic(nn.Module):
+	def __init__(self, state_dim, action_dim):
+		super(Critic, self).__init__()
+
+		self.l1 = nn.Linear(state_dim + action_dim, 400)
+		self.l2 = nn.Linear(400, 300)
+		self.l3 = nn.Linear(300, 1)
+
+
+	def forward(self, x, u):
+		x = F.relu(self.l1(torch.cat([x, u], 1)))
+		x = F.relu(self.l2(x))
+		x = self.l3(x)
+		return x 
+
+
+class DDPG(object):
+	def __init__(self, state_dim, action_dim, max_action):
+		self.actor = Actor(state_dim, action_dim, max_action)
+		self.actor_target = Actor(state_dim, action_dim, max_action)
+		self.actor_target.load_state_dict(self.actor.state_dict())
+		self.actor_optimizer = torch.optim.Adam(self.actor.parameters())
+
+		self.critic = Critic(state_dim, action_dim)
+		self.critic_target = Critic(state_dim, action_dim)
+		self.critic_target.load_state_dict(self.critic.state_dict())
+		self.critic_optimizer = torch.optim.Adam(self.critic.parameters())		
+
+		if torch.cuda.is_available():
+			self.actor = self.actor.cuda()
+			self.actor_target = self.actor_target.cuda()
+			self.critic = self.critic.cuda()
+			self.critic_target = self.critic_target.cuda()
+
+		self.criterion = nn.MSELoss()
+		self.state_dim = state_dim
+
+
+	def select_action(self, state):
+		state = var(torch.FloatTensor(state.reshape(-1, self.state_dim)), volatile=True)
+		return self.actor(state).cpu().data.numpy().flatten()
+
+
+	def train(self, replay_buffer, iterations, batch_size=100, discount=0.99, tau=0.005):
+
+		for it in range(iterations):
+
+			# Sample replay buffer 
+			x, y, u, r, d = replay_buffer.sample(batch_size)
+			state = var(torch.FloatTensor(x))
+			action = var(torch.FloatTensor(u))
+			next_state = var(torch.FloatTensor(y), volatile=True)
+			done = var(torch.FloatTensor(1 - d))
+			reward = var(torch.FloatTensor(r))
+
+			# Q target = reward + discount * Q(next_state, pi(next_state))
+			target_Q = self.critic_target(next_state, self.actor_target(next_state))
+			target_Q = reward + (done * discount * target_Q)
+			target_Q.volatile = False
+
+			# Get current Q estimate
+			current_Q = self.critic(state, action)
+
+			# Compute critic loss
+			critic_loss = self.criterion(current_Q, target_Q)
+
+			# Optimize the critic
+			self.critic_optimizer.zero_grad()
+			critic_loss.backward()
+			self.critic_optimizer.step()
+
+			# Compute actor loss
+			actor_loss = -self.critic(state, self.actor(state)).mean()
+
+			# Optimize the actor 
+			self.actor_optimizer.zero_grad()
+			actor_loss.backward()
+			self.actor_optimizer.step()
+
+			# Update the frozen target models
+			for param, target_param in zip(self.critic.parameters(), self.critic_target.parameters()):
+				target_param.data.copy_(tau * param.data + (1 - tau) * target_param.data)
+
+			for param, target_param in zip(self.actor.parameters(), self.actor_target.parameters()):
+				target_param.data.copy_(tau * param.data + (1 - tau) * target_param.data)
+
+
+	def save(self, filename, directory):
+		torch.save(self.actor.state_dict(), '%s/%s_actor.pth' % (directory, filename))
+		torch.save(self.critic.state_dict(), '%s/%s_critic.pth' % (directory, filename))
+
+
+	def load(self, filename, directory):
+		self.actor.load_state_dict(torch.load('%s/%s_actor.pth' % (directory, filename)))
+		self.critic.load_state_dict(torch.load('%s/%s_critic.pth' % (directory, filename)))

TD3.py

@@ -108,11 +108,11 @@ def train(self, replay_buffer, iterations, batch_size=100, discount=0.99, tau=0.
 			next_action = self.actor_target(next_state) + var(torch.FloatTensor(noise))
 			next_action = next_action.clamp(-self.max_action, self.max_action)
 
-			# Q target = reward + discount * min(Qi(next_state, pi(next_state)))
+			# Q target = reward + discount * min_i(Qi(next_state, pi(next_state)))
 			target_Q1, target_Q2 = self.critic_target(next_state, next_action)
-			target_Q = torch.min(torch.cat([target_Q1, target_Q2], 1), 1)[0].view(-1, 1)
-			target_Q.volatile = False 
+			target_Q = torch.min(target_Q1, target_Q2)
 			target_Q = reward + (done * discount * target_Q)
+			target_Q.volatile = False 
 
 			# Get current Q estimates
 			current_Q1, current_Q2 = self.critic(state, action)

main.py

@@ -5,8 +5,9 @@
 import os
 
 import utils
-import DDPG
 import TD3
+import OurDDPG
+import DDPG
 
 
 # Runs policy for X episodes and returns average reward
@@ -16,7 +17,7 @@ def evaluate_policy(policy, eval_episodes=10):
 		obs = env.reset()
 		done = False
 		while not done:
-			action = policy.select_action(np.array(obs)).clip(env.action_space.low, env.action_space.high)
+			action = policy.select_action(np.array(obs))
 			obs, reward, done, _ = env.step(action)
 			avg_reward += reward
 
@@ -69,8 +70,9 @@ def evaluate_policy(policy, eval_episodes=10):
 	max_action = int(env.action_space.high[0])
 
 	# Initialize policy
-	if args.policy_name == "DDPG": policy = DDPG.DDPG(state_dim, action_dim, max_action)
-	elif args.policy_name == "TD3": policy = TD3.TD3(state_dim, action_dim, max_action)
+	if args.policy_name == "TD3": policy = TD3.TD3(state_dim, action_dim, max_action)
+	elif args.policy_name == "OurDDPG": policy = OurDDPG.DDPG(state_dim, action_dim, max_action)
+	elif args.policy_name == "DDPG": policy = DDPG.DDPG(state_dim, action_dim, max_action)
 
 	replay_buffer = utils.ReplayBuffer()
 
@@ -98,7 +100,7 @@ def evaluate_policy(policy, eval_episodes=10):
 				timesteps_since_eval %= args.eval_freq
 				evaluations.append(evaluate_policy(policy))
 
-				if args.save_models: policy.save("%s" % (file_name), directory="./pytorch_models")
+				if args.save_models: policy.save(file_name, directory="./pytorch_models")
 				np.save("./results/%s" % (file_name), evaluations) 
 
 			# Reset environment