class Agent():
"""Interacts with and learns from the environment."""
def __init__(self, state_size, action_size, seed, mode="train"):
"""Initialize an Agent object.
Params
======
state_size (int): dimension of each state
action_size (int): dimension of each action
seed (int): random seed
mode (str): if eval, the agent will not learn and collect experiences
"""
self.state_size = state_size
self.action_size = action_size
self.seed = random.seed(seed)
# Q-Network
self.qnetwork_local = QNetwork(state_size, action_size,
seed).to(device)
self.qnetwork_target = QNetwork(state_size, action_size,
seed).to(device)
self.optimizer = optim.Adam(self.qnetwork_local.parameters(), lr=LR)
# Replay memory
self.memory = PrioritizedBuffer(action_size, BUFFER_SIZE, BATCH_SIZE,
seed)
# Initialize time step (for updating every UPDATE_EVERY steps)
self.t_step = 0
# Caches the expected action value of the last act
self.last_action_value = None
self.set_mode(mode)
def step(self, state, action, reward, next_state, done):
if self.mode == "train":
error = self.calculate_error_eval(state, action, reward,
next_state, done)
# Save experience in replay memory
self.memory.add(state, action, reward, next_state, done, error)
# Learn every UPDATE_EVERY time steps.
self.t_step = (self.t_step + 1) % UPDATE_EVERY
if self.t_step == 0:
# If enough samples are available in memory, get random subset and learn
if len(self.memory) > BATCH_SIZE:
experiences = self.memory.sample()
self.learn(experiences, GAMMA)
def act(self, state, eps=0.):
"""Returns actions for given state as per current policy.
Params
======
state (array_like): current state
eps (float): epsilon, for epsilon-greedy action selection
"""
if self.mode == "eval":
eps = 0
state = torch.from_numpy(state).float().unsqueeze(0).to(device)
self.qnetwork_local.eval()
with torch.no_grad():
action_values = self.qnetwork_local(state)
self.qnetwork_local.train()
# Epsilon-greedy action selection
if random.random() > eps:
action = np.argmax(action_values.cpu().data.numpy())
else:
action = random.choice(np.arange(self.action_size))
self.last_action_value = action_values[0][action].item()
return action
def learn(self, experiences, gamma):
"""Update value parameters using given batch of experience tuples.
Params
======
experiences (Tuple[torch.Variable]): tuple of (s, a, r, s', done) tuples
gamma (float): discount factor
"""
states, actions, rewards, next_states, dones = experiences
## TODO: compute and minimize the loss
with torch.no_grad():
local_max = self.qnetwork_local(next_states).detach().argmax(
1).unsqueeze(1)
targets = self.qnetwork_target(next_states).detach().gather(
1, local_max)
target_values = rewards + gamma * targets * (1 - dones)
predicted_values = self.qnetwork_local(states).gather(1, actions)
criterion = torch.nn.MSELoss()
loss = criterion(predicted_values, target_values)
self.optimizer.zero_grad()
loss.backward()
self.optimizer.step()
# ------------------- update target network ------------------- #
self.soft_update(self.qnetwork_local, self.qnetwork_target, TAU)
self.memory.update_errors((target_values - predicted_values).squeeze())
def soft_update(self, local_model, target_model, tau):
"""Soft update model parameters.
θ_target = τ*θ_local + (1 - τ)*θ_target
Params
======
local_model (PyTorch model): weights will be copied from
target_model (PyTorch model): weights will be copied to
tau (float): interpolation parameter
"""
for target_param, local_param in zip(target_model.parameters(),
local_model.parameters()):
target_param.data.copy_(tau * local_param.data +
(1.0 - tau) * target_param.data)
def calculate_error_eval(self, state, action, reward, next_state, done):
"""Calculates the error for a given step."""
self.qnetwork_target.eval()
next_state = torch.from_numpy(next_state).float().unsqueeze(0).to(
device)
with torch.no_grad():
target = self.qnetwork_target(next_state).max()
target_value = reward + GAMMA * target * (1 - done)
error = (target_value - self.last_action_value).item()
self.qnetwork_target.train()
return error
def save(self, path=""):
torch.save(self.qnetwork_local.state_dict(),
path + "checkpoint_local.pth")
torch.save(self.qnetwork_target.state_dict(),
path + "checkpoint_target.pth")
def load(self, path=""):
self.qnetwork_local.load_state_dict(
torch.load(path + "checkpoint_local.pth"))
self.qnetwork_target.load_state_dict(
torch.load(path + "checkpoint_target.pth"))
def set_mode(self, mode):
if mode not in {"train", "eval"}:
raise ValueError("mode must be one of [train, eval]")
self.mode = mode
class Agent():
"""Interacts with and learns from the environment."""
def __init__(self, state_size, action_size, seed
, local_filename = None, target_filename = None):
"""Initialize an Agent object.
Params
======
state_size (int): dimension of each state
action_size (int): dimension of each action
seed (int): random seed
"""
self.state_size = state_size
self.action_size = action_size
self.seed = random.seed(seed)
# Q-Network
self.qnetwork_local = QNetwork(state_size, action_size, seed).to(device)
self.qnetwork_target = QNetwork(state_size, action_size, seed).to(device)
## if filename is given load them
if local_filename is not None:
self.qnetwork_local.load_state_dict(torch.load(local_filename))
if target_filename is not None:
self.qnetwork_target.load_state_dict(torch.load(target_filename))
self.optimizer = optim.Adam(self.qnetwork_local.parameters(), lr=LR)
# Replay memory
self.memory = ReplayBuffer(action_size, BUFFER_SIZE, BATCH_SIZE, seed)
# Initialize time step (for updating every UPDATE_EVERY steps)
self.t_step = 0
def step(self, state, action, reward, next_state, done):
# Save experience in replay memory
self.memory.add(state, action, reward, next_state, done)
# Learn every UPDATE_EVERY time steps.
self.t_step = (self.t_step + 1) % UPDATE_EVERY
if self.t_step == 0:
# If enough samples are available in memory, get random subset and learn
if len(self.memory) > BATCH_SIZE:
experiences = self.memory.sample()
self.learn(experiences, GAMMA)
def act(self, state, eps=0.):
"""Returns actions for given state as per current policy.
Params
======
state (array_like): current state
eps (float): epsilon, for epsilon-greedy action selection
"""
state = torch.from_numpy(state).float().unsqueeze(0).to(device)
self.qnetwork_local.eval()
with torch.no_grad():
action_values = self.qnetwork_local(state)
self.qnetwork_local.train()
# Epsilon-greedy action selection
if random.random() > eps:
return np.argmax(action_values.cpu().data.numpy())
else:
return random.choice(np.arange(self.action_size))
def learn(self, experiences, gamma):
"""Update value parameters using given batch of experience tuples.
Params
======
experiences (Tuple[torch.Tensor]): tuple of (s, a, r, s', done) tuples
gamma (float): discount factor
"""
states, actions, rewards, next_states, dones = experiences
# Get max predicted Q values (for next states) from target model
Q_targets_next = self.qnetwork_target(next_states).detach().max(1)[0].unsqueeze(1)
# Compute Q targets for current states
Q_targets = rewards + (gamma * Q_targets_next * (1 - dones))
# Get expected Q values from local model
Q_expected = self.qnetwork_local(states).gather(1, actions)
# Compute loss
loss = F.mse_loss(Q_expected, Q_targets)
# Minimize the loss
self.optimizer.zero_grad()
loss.backward()
self.optimizer.step()
# ------------------- update target network ------------------- #
self.soft_update(self.qnetwork_local, self.qnetwork_target, TAU)
def soft_update(self, local_model, target_model, tau):
"""Soft update model parameters.
θ_target = τ*θ_local + (1 - τ)*θ_target
Params
======
local_model (PyTorch model): weights will be copied from
target_model (PyTorch model): weights will be copied to
tau (float): interpolation parameter
"""
for target_param, local_param in zip(target_model.parameters(), local_model.parameters()):
target_param.data.copy_(tau*local_param.data + (1.0-tau)*target_param.data)
def save(self, local_save_filename, target_save_filename):
torch.save(self.qnetwork_local.state_dict(), local_save_filename)
torch.save(self.qnetwork_target.state_dict(), target_save_filename)
