def appo_surrogate_loss(policy: Policy, model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch) -> TensorType:
"""Constructs the loss for APPO.
With IS modifications and V-trace for Advantage Estimation.
Args:
policy (Policy): The Policy to calculate the loss for.
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[ActionDistribution]: The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
model_out, _ = model.from_batch(train_batch)
action_dist = dist_class(model_out, model)
if isinstance(policy.action_space, gym.spaces.Discrete):
is_multidiscrete = False
output_hidden_shape = [policy.action_space.n]
elif isinstance(policy.action_space,
gym.spaces.multi_discrete.MultiDiscrete):
is_multidiscrete = True
output_hidden_shape = policy.action_space.nvec.astype(np.int32)
else:
is_multidiscrete = False
output_hidden_shape = 1
def _make_time_major(*args, **kw):
return make_time_major(policy, train_batch.get("seq_lens"), *args,
**kw)
actions = train_batch[SampleBatch.ACTIONS]
dones = train_batch[SampleBatch.DONES]
rewards = train_batch[SampleBatch.REWARDS]
behaviour_logits = train_batch[SampleBatch.ACTION_DIST_INPUTS]
target_model_out, _ = policy.target_model.from_batch(train_batch)
prev_action_dist = dist_class(behaviour_logits, policy.model)
values = policy.model.value_function()
values_time_major = _make_time_major(values)
policy.model_vars = policy.model.variables()
policy.target_model_vars = policy.target_model.variables()
if policy.is_recurrent():
max_seq_len = torch.max(train_batch["seq_lens"]) - 1
mask = sequence_mask(train_batch["seq_lens"], max_seq_len)
mask = torch.reshape(mask, [-1])
num_valid = torch.sum(mask)
def reduce_mean_valid(t):
return torch.sum(t * mask) / num_valid
else:
reduce_mean_valid = torch.mean
if policy.config["vtrace"]:
logger.debug("Using V-Trace surrogate loss (vtrace=True)")
old_policy_behaviour_logits = target_model_out.detach()
old_policy_action_dist = dist_class(old_policy_behaviour_logits, model)
if isinstance(output_hidden_shape, (list, tuple, np.ndarray)):
unpacked_behaviour_logits = torch.split(behaviour_logits,
list(output_hidden_shape),
dim=1)
unpacked_old_policy_behaviour_logits = torch.split(
old_policy_behaviour_logits, list(output_hidden_shape), dim=1)
else:
unpacked_behaviour_logits = torch.chunk(behaviour_logits,
output_hidden_shape,
dim=1)
unpacked_old_policy_behaviour_logits = torch.chunk(
old_policy_behaviour_logits, output_hidden_shape, dim=1)
# Prepare actions for loss
loss_actions = actions if is_multidiscrete else torch.unsqueeze(
actions, dim=1)
# Prepare KL for Loss
action_kl = _make_time_major(old_policy_action_dist.kl(action_dist),
drop_last=True)
# Compute vtrace on the CPU for better perf.
vtrace_returns = vtrace.multi_from_logits(
behaviour_policy_logits=_make_time_major(unpacked_behaviour_logits,
drop_last=True),
target_policy_logits=_make_time_major(
unpacked_old_policy_behaviour_logits, drop_last=True),
actions=torch.unbind(_make_time_major(loss_actions,
drop_last=True),
dim=2),
discounts=(1.0 - _make_time_major(dones, drop_last=True).float()) *
policy.config["gamma"],
rewards=_make_time_major(rewards, drop_last=True),
values=values_time_major[:-1], # drop-last=True
bootstrap_value=values_time_major[-1],
dist_class=TorchCategorical if is_multidiscrete else dist_class,
model=model,
clip_rho_threshold=policy.config["vtrace_clip_rho_threshold"],
clip_pg_rho_threshold=policy.config["vtrace_clip_pg_rho_threshold"]
)
actions_logp = _make_time_major(action_dist.logp(actions),
drop_last=True)
prev_actions_logp = _make_time_major(prev_action_dist.logp(actions),
drop_last=True)
old_policy_actions_logp = _make_time_major(
old_policy_action_dist.logp(actions), drop_last=True)
is_ratio = torch.clamp(
torch.exp(prev_actions_logp - old_policy_actions_logp), 0.0, 2.0)
logp_ratio = is_ratio * torch.exp(actions_logp - prev_actions_logp)
policy._is_ratio = is_ratio
advantages = vtrace_returns.pg_advantages
surrogate_loss = torch.min(
advantages * logp_ratio,
advantages *
torch.clamp(logp_ratio, 1 - policy.config["clip_param"],
1 + policy.config["clip_param"]))
mean_kl = reduce_mean_valid(action_kl)
mean_policy_loss = -reduce_mean_valid(surrogate_loss)
# The value function loss.
delta = values_time_major[:-1] - vtrace_returns.vs
value_targets = vtrace_returns.vs
mean_vf_loss = 0.5 * reduce_mean_valid(torch.pow(delta, 2.0))
# The entropy loss.
mean_entropy = reduce_mean_valid(
_make_time_major(action_dist.entropy(), drop_last=True))
else:
logger.debug("Using PPO surrogate loss (vtrace=False)")
# Prepare KL for Loss
action_kl = _make_time_major(prev_action_dist.kl(action_dist))
actions_logp = _make_time_major(action_dist.logp(actions))
prev_actions_logp = _make_time_major(prev_action_dist.logp(actions))
logp_ratio = torch.exp(actions_logp - prev_actions_logp)
advantages = _make_time_major(train_batch[Postprocessing.ADVANTAGES])
surrogate_loss = torch.min(
advantages * logp_ratio,
advantages *
torch.clamp(logp_ratio, 1 - policy.config["clip_param"],
1 + policy.config["clip_param"]))
mean_kl = reduce_mean_valid(action_kl)
mean_policy_loss = -reduce_mean_valid(surrogate_loss)
# The value function loss.
value_targets = _make_time_major(
train_batch[Postprocessing.VALUE_TARGETS])
delta = values_time_major - value_targets
mean_vf_loss = 0.5 * reduce_mean_valid(torch.pow(delta, 2.0))
# The entropy loss.
mean_entropy = reduce_mean_valid(
_make_time_major(action_dist.entropy()))
# The summed weighted loss
total_loss = mean_policy_loss + \
mean_vf_loss * policy.config["vf_loss_coeff"] - \
mean_entropy * policy.config["entropy_coeff"]
# Optional additional KL Loss
if policy.config["use_kl_loss"]:
total_loss += policy.kl_coeff * mean_kl
policy._total_loss = total_loss
policy._mean_policy_loss = mean_policy_loss
policy._mean_kl = mean_kl
policy._mean_vf_loss = mean_vf_loss
policy._mean_entropy = mean_entropy
policy._value_targets = value_targets
return total_loss
def ppo_surrogate_loss(
policy: Policy, model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss for Proximal Policy Objective.
Args:
policy (Policy): The Policy to calculate the loss for.
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[ActionDistribution]: The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
logits, state = model.from_batch(train_batch, is_training=True)
curr_action_dist = dist_class(logits, model)
# RNN case: Mask away 0-padded chunks at end of time axis.
if state:
max_seq_len = torch.max(train_batch["seq_lens"])
mask = sequence_mask(train_batch["seq_lens"],
max_seq_len,
time_major=model.is_time_major())
mask = torch.reshape(mask, [-1])
num_valid = torch.sum(mask)
def reduce_mean_valid(t):
return torch.sum(t[mask]) / num_valid
# non-RNN case: No masking.
else:
mask = None
reduce_mean_valid = torch.mean
prev_action_dist = dist_class(train_batch[SampleBatch.ACTION_DIST_INPUTS],
model)
logp_ratio = torch.exp(
curr_action_dist.logp(train_batch[SampleBatch.ACTIONS]) -
train_batch[SampleBatch.ACTION_LOGP])
action_kl = prev_action_dist.kl(curr_action_dist)
mean_kl = reduce_mean_valid(action_kl)
curr_entropy = curr_action_dist.entropy()
mean_entropy = reduce_mean_valid(curr_entropy)
surrogate_loss = torch.min(
train_batch[Postprocessing.ADVANTAGES] * logp_ratio,
train_batch[Postprocessing.ADVANTAGES] *
torch.clamp(logp_ratio, 1 - policy.config["clip_param"],
1 + policy.config["clip_param"]))
mean_policy_loss = reduce_mean_valid(-surrogate_loss)
if policy.config["use_gae"]:
prev_value_fn_out = train_batch[SampleBatch.VF_PREDS]
value_fn_out = model.value_function()
vf_loss1 = torch.pow(
value_fn_out - train_batch[Postprocessing.VALUE_TARGETS], 2.0)
vf_clipped = prev_value_fn_out + torch.clamp(
value_fn_out - prev_value_fn_out, -policy.config["vf_clip_param"],
policy.config["vf_clip_param"])
vf_loss2 = torch.pow(
vf_clipped - train_batch[Postprocessing.VALUE_TARGETS], 2.0)
vf_loss = torch.max(vf_loss1, vf_loss2)
mean_vf_loss = reduce_mean_valid(vf_loss)
total_loss = reduce_mean_valid(-surrogate_loss +
policy.kl_coeff * action_kl +
policy.config["vf_loss_coeff"] *
vf_loss -
policy.entropy_coeff * curr_entropy)
else:
mean_vf_loss = 0.0
total_loss = reduce_mean_valid(-surrogate_loss +
policy.kl_coeff * action_kl -
policy.entropy_coeff * curr_entropy)
# Store stats in policy for stats_fn.
policy._total_loss = total_loss
policy._mean_policy_loss = mean_policy_loss
policy._mean_vf_loss = mean_vf_loss
policy._mean_entropy = mean_entropy
policy._mean_kl = mean_kl
return total_loss
def cql_loss(
policy: Policy,
model: ModelV2,
dist_class: Type[TFActionDistribution],
train_batch: SampleBatch,
) -> Union[TensorType, List[TensorType]]:
logger.info(f"Current iteration = {policy.cur_iter}")
policy.cur_iter += 1
# For best performance, turn deterministic off
deterministic = policy.config["_deterministic_loss"]
assert not deterministic
twin_q = policy.config["twin_q"]
discount = policy.config["gamma"]
# CQL Parameters
bc_iters = policy.config["bc_iters"]
cql_temp = policy.config["temperature"]
num_actions = policy.config["num_actions"]
min_q_weight = policy.config["min_q_weight"]
use_lagrange = policy.config["lagrangian"]
target_action_gap = policy.config["lagrangian_thresh"]
obs = train_batch[SampleBatch.CUR_OBS]
actions = tf.cast(train_batch[SampleBatch.ACTIONS], tf.float32)
rewards = tf.cast(train_batch[SampleBatch.REWARDS], tf.float32)
next_obs = train_batch[SampleBatch.NEXT_OBS]
terminals = train_batch[SampleBatch.DONES]
model_out_t, _ = model(SampleBatch(obs=obs, _is_training=True), [], None)
model_out_tp1, _ = model(SampleBatch(obs=next_obs, _is_training=True), [],
None)
target_model_out_tp1, _ = policy.target_model(
SampleBatch(obs=next_obs, _is_training=True), [], None)
action_dist_class = _get_dist_class(policy, policy.config,
policy.action_space)
action_dist_inputs_t, _ = model.get_action_model_outputs(model_out_t)
action_dist_t = action_dist_class(action_dist_inputs_t, model)
policy_t, log_pis_t = action_dist_t.sample_logp()
log_pis_t = tf.expand_dims(log_pis_t, -1)
# Unlike original SAC, Alpha and Actor Loss are computed first.
# Alpha Loss
alpha_loss = -tf.reduce_mean(
model.log_alpha * tf.stop_gradient(log_pis_t + model.target_entropy))
# Policy Loss (Either Behavior Clone Loss or SAC Loss)
alpha = tf.math.exp(model.log_alpha)
if policy.cur_iter >= bc_iters:
min_q, _ = model.get_q_values(model_out_t, policy_t)
if twin_q:
twin_q_, _ = model.get_twin_q_values(model_out_t, policy_t)
min_q = tf.math.minimum(min_q, twin_q_)
actor_loss = tf.reduce_mean(
tf.stop_gradient(alpha) * log_pis_t - min_q)
else:
bc_logp = action_dist_t.logp(actions)
actor_loss = tf.reduce_mean(
tf.stop_gradient(alpha) * log_pis_t - bc_logp)
# actor_loss = -tf.reduce_mean(bc_logp)
# Critic Loss (Standard SAC Critic L2 Loss + CQL Entropy Loss)
# SAC Loss:
# Q-values for the batched actions.
action_dist_inputs_tp1, _ = model.get_action_model_outputs(model_out_tp1)
action_dist_tp1 = action_dist_class(action_dist_inputs_tp1, model)
policy_tp1, _ = action_dist_tp1.sample_logp()
q_t, _ = model.get_q_values(model_out_t, actions)
q_t_selected = tf.squeeze(q_t, axis=-1)
if twin_q:
twin_q_t, _ = model.get_twin_q_values(model_out_t, actions)
twin_q_t_selected = tf.squeeze(twin_q_t, axis=-1)
# Target q network evaluation.
q_tp1, _ = policy.target_model.get_q_values(target_model_out_tp1,
policy_tp1)
if twin_q:
twin_q_tp1, _ = policy.target_model.get_twin_q_values(
target_model_out_tp1, policy_tp1)
# Take min over both twin-NNs.
q_tp1 = tf.math.minimum(q_tp1, twin_q_tp1)
q_tp1_best = tf.squeeze(input=q_tp1, axis=-1)
q_tp1_best_masked = (1.0 - tf.cast(terminals, tf.float32)) * q_tp1_best
# compute RHS of bellman equation
q_t_target = tf.stop_gradient(rewards +
(discount**policy.config["n_step"]) *
q_tp1_best_masked)
# Compute the TD-error (potentially clipped), for priority replay buffer
base_td_error = tf.math.abs(q_t_selected - q_t_target)
if twin_q:
twin_td_error = tf.math.abs(twin_q_t_selected - q_t_target)
td_error = 0.5 * (base_td_error + twin_td_error)
else:
td_error = base_td_error
critic_loss_1 = tf.keras.losses.MSE(q_t_selected, q_t_target)
if twin_q:
critic_loss_2 = tf.keras.losses.MSE(twin_q_t_selected, q_t_target)
# CQL Loss (We are using Entropy version of CQL (the best version))
rand_actions, _ = policy._random_action_generator.get_exploration_action(
action_distribution=action_dist_class(
tf.tile(action_dist_tp1.inputs, (num_actions, 1)), model),
timestep=0,
explore=True,
)
curr_actions, curr_logp = policy_actions_repeat(model, action_dist_class,
model_out_t, num_actions)
next_actions, next_logp = policy_actions_repeat(model, action_dist_class,
model_out_tp1, num_actions)
q1_rand = q_values_repeat(model, model_out_t, rand_actions)
q1_curr_actions = q_values_repeat(model, model_out_t, curr_actions)
q1_next_actions = q_values_repeat(model, model_out_t, next_actions)
if twin_q:
q2_rand = q_values_repeat(model, model_out_t, rand_actions, twin=True)
q2_curr_actions = q_values_repeat(model,
model_out_t,
curr_actions,
twin=True)
q2_next_actions = q_values_repeat(model,
model_out_t,
next_actions,
twin=True)
random_density = np.log(0.5**int(curr_actions.shape[-1]))
cat_q1 = tf.concat(
[
q1_rand - random_density,
q1_next_actions - tf.stop_gradient(next_logp),
q1_curr_actions - tf.stop_gradient(curr_logp),
],
1,
)
if twin_q:
cat_q2 = tf.concat(
[
q2_rand - random_density,
q2_next_actions - tf.stop_gradient(next_logp),
q2_curr_actions - tf.stop_gradient(curr_logp),
],
1,
)
min_qf1_loss_ = (
tf.reduce_mean(tf.reduce_logsumexp(cat_q1 / cql_temp, axis=1)) *
min_q_weight * cql_temp)
min_qf1_loss = min_qf1_loss_ - (tf.reduce_mean(q_t) * min_q_weight)
if twin_q:
min_qf2_loss_ = (
tf.reduce_mean(tf.reduce_logsumexp(cat_q2 / cql_temp, axis=1)) *
min_q_weight * cql_temp)
min_qf2_loss = min_qf2_loss_ - (tf.reduce_mean(twin_q_t) *
min_q_weight)
if use_lagrange:
alpha_prime = tf.clip_by_value(model.log_alpha_prime.exp(), 0.0,
1000000.0)[0]
min_qf1_loss = alpha_prime * (min_qf1_loss - target_action_gap)
if twin_q:
min_qf2_loss = alpha_prime * (min_qf2_loss - target_action_gap)
alpha_prime_loss = 0.5 * (-min_qf1_loss - min_qf2_loss)
else:
alpha_prime_loss = -min_qf1_loss
cql_loss = [min_qf1_loss]
if twin_q:
cql_loss.append(min_qf2_loss)
critic_loss = [critic_loss_1 + min_qf1_loss]
if twin_q:
critic_loss.append(critic_loss_2 + min_qf2_loss)
# Save for stats function.
policy.q_t = q_t_selected
policy.policy_t = policy_t
policy.log_pis_t = log_pis_t
policy.td_error = td_error
policy.actor_loss = actor_loss
policy.critic_loss = critic_loss
policy.alpha_loss = alpha_loss
policy.log_alpha_value = model.log_alpha
policy.alpha_value = alpha
policy.target_entropy = model.target_entropy
# CQL Stats
policy.cql_loss = cql_loss
if use_lagrange:
policy.log_alpha_prime_value = model.log_alpha_prime[0]
policy.alpha_prime_value = alpha_prime
policy.alpha_prime_loss = alpha_prime_loss
# Return all loss terms corresponding to our optimizers.
if use_lagrange:
return actor_loss + tf.math.add_n(
critic_loss) + alpha_loss + alpha_prime_loss
return actor_loss + tf.math.add_n(critic_loss) + alpha_loss
def context(self):
"""Returns a contextmanager for the current TF graph."""
if self.graph:
return self.graph.as_default()
else:
return ModelV2.context(self)
def actor_critic_loss(
policy: Policy, model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss for the Soft Actor Critic.
Args:
policy (Policy): The Policy to calculate the loss for.
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[TorchDistributionWrapper]: The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
# Look up the target model (tower) using the model tower.
target_model = policy.target_models[model]
# Should be True only for debugging purposes (e.g. test cases)!
deterministic = policy.config["_deterministic_loss"]
model_out_t, _ = model(
{
"obs": train_batch[SampleBatch.CUR_OBS],
"is_training": True,
}, [], None)
model_out_tp1, _ = model(
{
"obs": train_batch[SampleBatch.NEXT_OBS],
"is_training": True,
}, [], None)
target_model_out_tp1, _ = target_model(
{
"obs": train_batch[SampleBatch.NEXT_OBS],
"is_training": True,
}, [], None)
alpha = torch.exp(model.log_alpha)
# Discrete case.
if model.discrete:
# Get all action probs directly from pi and form their logp.
log_pis_t = F.log_softmax(model.get_policy_output(model_out_t), dim=-1)
policy_t = torch.exp(log_pis_t)
log_pis_tp1 = F.log_softmax(model.get_policy_output(model_out_tp1), -1)
policy_tp1 = torch.exp(log_pis_tp1)
# Q-values.
q_t = model.get_q_values(model_out_t)
# Target Q-values.
q_tp1 = target_model.get_q_values(target_model_out_tp1)
if policy.config["twin_q"]:
twin_q_t = model.get_twin_q_values(model_out_t)
twin_q_tp1 = target_model.get_twin_q_values(target_model_out_tp1)
q_tp1 = torch.min(q_tp1, twin_q_tp1)
q_tp1 -= alpha * log_pis_tp1
# Actually selected Q-values (from the actions batch).
one_hot = F.one_hot(train_batch[SampleBatch.ACTIONS].long(),
num_classes=q_t.size()[-1])
q_t_selected = torch.sum(q_t * one_hot, dim=-1)
if policy.config["twin_q"]:
twin_q_t_selected = torch.sum(twin_q_t * one_hot, dim=-1)
# Discrete case: "Best" means weighted by the policy (prob) outputs.
q_tp1_best = torch.sum(torch.mul(policy_tp1, q_tp1), dim=-1)
q_tp1_best_masked = \
(1.0 - train_batch[SampleBatch.DONES].float()) * \
q_tp1_best
# Continuous actions case.
else:
# Sample single actions from distribution.
action_dist_class = _get_dist_class(policy, policy.config,
policy.action_space)
action_dist_t = action_dist_class(model.get_policy_output(model_out_t),
model)
policy_t = action_dist_t.sample() if not deterministic else \
action_dist_t.deterministic_sample()
log_pis_t = torch.unsqueeze(action_dist_t.logp(policy_t), -1)
action_dist_tp1 = action_dist_class(
model.get_policy_output(model_out_tp1), model)
policy_tp1 = action_dist_tp1.sample() if not deterministic else \
action_dist_tp1.deterministic_sample()
log_pis_tp1 = torch.unsqueeze(action_dist_tp1.logp(policy_tp1), -1)
# Q-values for the actually selected actions.
q_t = model.get_q_values(model_out_t, train_batch[SampleBatch.ACTIONS])
if policy.config["twin_q"]:
twin_q_t = model.get_twin_q_values(
model_out_t, train_batch[SampleBatch.ACTIONS])
# Q-values for current policy in given current state.
q_t_det_policy = model.get_q_values(model_out_t, policy_t)
if policy.config["twin_q"]:
twin_q_t_det_policy = model.get_twin_q_values(
model_out_t, policy_t)
q_t_det_policy = torch.min(q_t_det_policy, twin_q_t_det_policy)
# Target q network evaluation.
q_tp1 = target_model.get_q_values(target_model_out_tp1, policy_tp1)
if policy.config["twin_q"]:
twin_q_tp1 = target_model.get_twin_q_values(
target_model_out_tp1, policy_tp1)
# Take min over both twin-NNs.
q_tp1 = torch.min(q_tp1, twin_q_tp1)
q_t_selected = torch.squeeze(q_t, dim=-1)
if policy.config["twin_q"]:
twin_q_t_selected = torch.squeeze(twin_q_t, dim=-1)
q_tp1 -= alpha * log_pis_tp1
q_tp1_best = torch.squeeze(input=q_tp1, dim=-1)
q_tp1_best_masked = (1.0 - train_batch[SampleBatch.DONES].float()) * \
q_tp1_best
# compute RHS of bellman equation
q_t_selected_target = (train_batch[SampleBatch.REWARDS] +
(policy.config["gamma"]**policy.config["n_step"]) *
q_tp1_best_masked).detach()
# Compute the TD-error (potentially clipped).
base_td_error = torch.abs(q_t_selected - q_t_selected_target)
if policy.config["twin_q"]:
twin_td_error = torch.abs(twin_q_t_selected - q_t_selected_target)
td_error = 0.5 * (base_td_error + twin_td_error)
else:
td_error = base_td_error
critic_loss = [
torch.mean(train_batch[PRIO_WEIGHTS] * huber_loss(base_td_error))
]
if policy.config["twin_q"]:
critic_loss.append(
torch.mean(train_batch[PRIO_WEIGHTS] * huber_loss(twin_td_error)))
# Alpha- and actor losses.
# Note: In the papers, alpha is used directly, here we take the log.
# Discrete case: Multiply the action probs as weights with the original
# loss terms (no expectations needed).
if model.discrete:
weighted_log_alpha_loss = policy_t.detach() * (
-model.log_alpha * (log_pis_t + model.target_entropy).detach())
# Sum up weighted terms and mean over all batch items.
alpha_loss = torch.mean(torch.sum(weighted_log_alpha_loss, dim=-1))
# Actor loss.
actor_loss = torch.mean(
torch.sum(
torch.mul(
# NOTE: No stop_grad around policy output here
# (compare with q_t_det_policy for continuous case).
policy_t,
alpha.detach() * log_pis_t - q_t.detach()),
dim=-1))
else:
alpha_loss = -torch.mean(model.log_alpha *
(log_pis_t + model.target_entropy).detach())
# Note: Do not detach q_t_det_policy here b/c is depends partly
# on the policy vars (policy sample pushed through Q-net).
# However, we must make sure `actor_loss` is not used to update
# the Q-net(s)' variables.
actor_loss = torch.mean(alpha.detach() * log_pis_t - q_t_det_policy)
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["q_t"] = q_t
model.tower_stats["policy_t"] = policy_t
model.tower_stats["log_pis_t"] = log_pis_t
model.tower_stats["actor_loss"] = actor_loss
model.tower_stats["critic_loss"] = critic_loss
model.tower_stats["alpha_loss"] = alpha_loss
# TD-error tensor in final stats
# will be concatenated and retrieved for each individual batch item.
model.tower_stats["td_error"] = td_error
# Return all loss terms corresponding to our optimizers.
return tuple([actor_loss] + critic_loss + [alpha_loss])
def cql_loss(policy: Policy, model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch) -> Union[TensorType, List[TensorType]]:
logger.info(f"Current iteration = {policy.cur_iter}")
policy.cur_iter += 1
# Look up the target model (tower) using the model tower.
target_model = policy.target_models[model]
# For best performance, turn deterministic off
deterministic = policy.config["_deterministic_loss"]
assert not deterministic
twin_q = policy.config["twin_q"]
discount = policy.config["gamma"]
action_low = model.action_space.low[0]
action_high = model.action_space.high[0]
# CQL Parameters
bc_iters = policy.config["bc_iters"]
cql_temp = policy.config["temperature"]
num_actions = policy.config["num_actions"]
min_q_weight = policy.config["min_q_weight"]
use_lagrange = policy.config["lagrangian"]
target_action_gap = policy.config["lagrangian_thresh"]
obs = train_batch[SampleBatch.CUR_OBS]
actions = train_batch[SampleBatch.ACTIONS]
rewards = train_batch[SampleBatch.REWARDS].float()
next_obs = train_batch[SampleBatch.NEXT_OBS]
terminals = train_batch[SampleBatch.DONES]
model_out_t, _ = model({
"obs": obs,
"is_training": True,
}, [], None)
model_out_tp1, _ = model({
"obs": next_obs,
"is_training": True,
}, [], None)
target_model_out_tp1, _ = target_model({
"obs": next_obs,
"is_training": True,
}, [], None)
action_dist_class = _get_dist_class(policy, policy.config,
policy.action_space)
action_dist_t = action_dist_class(
model.get_policy_output(model_out_t), policy.model)
policy_t, log_pis_t = action_dist_t.sample_logp()
log_pis_t = torch.unsqueeze(log_pis_t, -1)
# Unlike original SAC, Alpha and Actor Loss are computed first.
# Alpha Loss
alpha_loss = -(model.log_alpha *
(log_pis_t + model.target_entropy).detach()).mean()
if obs.shape[0] == policy.config["train_batch_size"]:
policy.alpha_optim.zero_grad()
alpha_loss.backward()
policy.alpha_optim.step()
# Policy Loss (Either Behavior Clone Loss or SAC Loss)
alpha = torch.exp(model.log_alpha)
if policy.cur_iter >= bc_iters:
min_q = model.get_q_values(model_out_t, policy_t)
if twin_q:
twin_q_ = model.get_twin_q_values(model_out_t, policy_t)
min_q = torch.min(min_q, twin_q_)
actor_loss = (alpha.detach() * log_pis_t - min_q).mean()
else:
bc_logp = action_dist_t.logp(actions)
actor_loss = (alpha.detach() * log_pis_t - bc_logp).mean()
# actor_loss = -bc_logp.mean()
if obs.shape[0] == policy.config["train_batch_size"]:
policy.actor_optim.zero_grad()
actor_loss.backward(retain_graph=True)
policy.actor_optim.step()
# Critic Loss (Standard SAC Critic L2 Loss + CQL Entropy Loss)
# SAC Loss:
# Q-values for the batched actions.
action_dist_tp1 = action_dist_class(
model.get_policy_output(model_out_tp1), policy.model)
policy_tp1, _ = action_dist_tp1.sample_logp()
q_t = model.get_q_values(model_out_t, train_batch[SampleBatch.ACTIONS])
q_t_selected = torch.squeeze(q_t, dim=-1)
if twin_q:
twin_q_t = model.get_twin_q_values(model_out_t,
train_batch[SampleBatch.ACTIONS])
twin_q_t_selected = torch.squeeze(twin_q_t, dim=-1)
# Target q network evaluation.
q_tp1 = target_model.get_q_values(target_model_out_tp1, policy_tp1)
if twin_q:
twin_q_tp1 = target_model.get_twin_q_values(target_model_out_tp1,
policy_tp1)
# Take min over both twin-NNs.
q_tp1 = torch.min(q_tp1, twin_q_tp1)
q_tp1_best = torch.squeeze(input=q_tp1, dim=-1)
q_tp1_best_masked = (1.0 - terminals.float()) * q_tp1_best
# compute RHS of bellman equation
q_t_target = (
rewards +
(discount**policy.config["n_step"]) * q_tp1_best_masked).detach()
# Compute the TD-error (potentially clipped), for priority replay buffer
base_td_error = torch.abs(q_t_selected - q_t_target)
if twin_q:
twin_td_error = torch.abs(twin_q_t_selected - q_t_target)
td_error = 0.5 * (base_td_error + twin_td_error)
else:
td_error = base_td_error
critic_loss_1 = nn.functional.mse_loss(q_t_selected, q_t_target)
if twin_q:
critic_loss_2 = nn.functional.mse_loss(twin_q_t_selected, q_t_target)
# CQL Loss (We are using Entropy version of CQL (the best version))
rand_actions = convert_to_torch_tensor(
torch.FloatTensor(actions.shape[0] * num_actions,
actions.shape[-1]).uniform_(action_low, action_high),
policy.device)
curr_actions, curr_logp = policy_actions_repeat(model, action_dist_class,
model_out_t, num_actions)
next_actions, next_logp = policy_actions_repeat(model, action_dist_class,
model_out_tp1, num_actions)
q1_rand = q_values_repeat(model, model_out_t, rand_actions)
q1_curr_actions = q_values_repeat(model, model_out_t, curr_actions)
q1_next_actions = q_values_repeat(model, model_out_t, next_actions)
if twin_q:
q2_rand = q_values_repeat(model, model_out_t, rand_actions, twin=True)
q2_curr_actions = q_values_repeat(
model, model_out_t, curr_actions, twin=True)
q2_next_actions = q_values_repeat(
model, model_out_t, next_actions, twin=True)
random_density = np.log(0.5**curr_actions.shape[-1])
cat_q1 = torch.cat([
q1_rand - random_density, q1_next_actions - next_logp.detach(),
q1_curr_actions - curr_logp.detach()
], 1)
if twin_q:
cat_q2 = torch.cat([
q2_rand - random_density, q2_next_actions - next_logp.detach(),
q2_curr_actions - curr_logp.detach()
], 1)
min_qf1_loss_ = torch.logsumexp(
cat_q1 / cql_temp, dim=1).mean() * min_q_weight * cql_temp
min_qf1_loss = min_qf1_loss_ - (q_t.mean() * min_q_weight)
if twin_q:
min_qf2_loss_ = torch.logsumexp(
cat_q2 / cql_temp, dim=1).mean() * min_q_weight * cql_temp
min_qf2_loss = min_qf2_loss_ - (twin_q_t.mean() * min_q_weight)
if use_lagrange:
alpha_prime = torch.clamp(
model.log_alpha_prime.exp(), min=0.0, max=1000000.0)[0]
min_qf1_loss = alpha_prime * (min_qf1_loss - target_action_gap)
if twin_q:
min_qf2_loss = alpha_prime * (min_qf2_loss - target_action_gap)
alpha_prime_loss = 0.5 * (-min_qf1_loss - min_qf2_loss)
else:
alpha_prime_loss = -min_qf1_loss
cql_loss = [min_qf1_loss]
if twin_q:
cql_loss.append(min_qf2_loss)
critic_loss = [critic_loss_1 + min_qf1_loss]
if twin_q:
critic_loss.append(critic_loss_2 + min_qf2_loss)
if obs.shape[0] == policy.config["train_batch_size"]:
policy.critic_optims[0].zero_grad()
critic_loss[0].backward(retain_graph=True)
policy.critic_optims[0].step()
if twin_q:
policy.critic_optims[1].zero_grad()
critic_loss[1].backward(retain_graph=False)
policy.critic_optims[1].step()
# Save for stats function.
policy.q_t = q_t_selected
policy.policy_t = policy_t
policy.log_pis_t = log_pis_t
model.td_error = td_error
policy.actor_loss = actor_loss
policy.critic_loss = critic_loss
policy.alpha_loss = alpha_loss
policy.log_alpha_value = model.log_alpha
policy.alpha_value = alpha
policy.target_entropy = model.target_entropy
# CQL Stats.
policy.cql_loss = cql_loss
if use_lagrange:
policy.log_alpha_prime_value = model.log_alpha_prime[0]
policy.alpha_prime_value = alpha_prime
policy.alpha_prime_loss = alpha_prime_loss
if obs.shape[0] == policy.config["train_batch_size"]:
policy.alpha_prime_optim.zero_grad()
alpha_prime_loss.backward()
policy.alpha_prime_optim.step()
# Return all loss terms corresponding to our optimizers.
if use_lagrange:
return tuple([policy.actor_loss] + policy.critic_loss +
[policy.alpha_loss] + [policy.alpha_prime_loss])
return tuple([policy.actor_loss] + policy.critic_loss +
[policy.alpha_loss])
def loss(
self,
model: ModelV2,
dist_class: Type[ActionDistribution],
train_batch: SampleBatch,
) -> Union[TensorType, List[TensorType]]:
"""Compute loss for Proximal Policy Objective.
Args:
model: The Model to calculate the loss for.
dist_class: The action distr. class.
train_batch: The training data.
Returns:
The PPO loss tensor given the input batch.
"""
logits, state = model(train_batch)
curr_action_dist = dist_class(logits, model)
# RNN case: Mask away 0-padded chunks at end of time axis.
if state:
B = len(train_batch[SampleBatch.SEQ_LENS])
max_seq_len = logits.shape[0] // B
mask = sequence_mask(
train_batch[SampleBatch.SEQ_LENS],
max_seq_len,
time_major=model.is_time_major(),
)
mask = torch.reshape(mask, [-1])
num_valid = torch.sum(mask)
def reduce_mean_valid(t):
return torch.sum(t[mask]) / num_valid
# non-RNN case: No masking.
else:
mask = None
reduce_mean_valid = torch.mean
prev_action_dist = dist_class(
train_batch[SampleBatch.ACTION_DIST_INPUTS], model)
logp_ratio = torch.exp(
curr_action_dist.logp(train_batch[SampleBatch.ACTIONS]) -
train_batch[SampleBatch.ACTION_LOGP])
# Only calculate kl loss if necessary (kl-coeff > 0.0).
if self.config["kl_coeff"] > 0.0:
action_kl = prev_action_dist.kl(curr_action_dist)
mean_kl_loss = reduce_mean_valid(action_kl)
else:
mean_kl_loss = torch.tensor(0.0, device=logp_ratio.device)
curr_entropy = curr_action_dist.entropy()
mean_entropy = reduce_mean_valid(curr_entropy)
surrogate_loss = torch.min(
train_batch[Postprocessing.ADVANTAGES] * logp_ratio,
train_batch[Postprocessing.ADVANTAGES] *
torch.clamp(logp_ratio, 1 - self.config["clip_param"],
1 + self.config["clip_param"]),
)
mean_policy_loss = reduce_mean_valid(-surrogate_loss)
# Compute a value function loss.
if self.config["use_critic"]:
value_fn_out = model.value_function()
vf_loss = torch.pow(
value_fn_out - train_batch[Postprocessing.VALUE_TARGETS], 2.0)
vf_loss_clipped = torch.clamp(vf_loss, 0,
self.config["vf_clip_param"])
mean_vf_loss = reduce_mean_valid(vf_loss_clipped)
# Ignore the value function.
else:
value_fn_out = 0
vf_loss_clipped = mean_vf_loss = 0.0
total_loss = reduce_mean_valid(-surrogate_loss +
self.config["vf_loss_coeff"] *
vf_loss_clipped -
self.entropy_coeff * curr_entropy)
# Add mean_kl_loss (already processed through `reduce_mean_valid`),
# if necessary.
if self.config["kl_coeff"] > 0.0:
total_loss += self.kl_coeff * mean_kl_loss
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["total_loss"] = total_loss
model.tower_stats["mean_policy_loss"] = mean_policy_loss
model.tower_stats["mean_vf_loss"] = mean_vf_loss
model.tower_stats["vf_explained_var"] = explained_variance(
train_batch[Postprocessing.VALUE_TARGETS], value_fn_out)
model.tower_stats["mean_entropy"] = mean_entropy
model.tower_stats["mean_kl_loss"] = mean_kl_loss
return total_loss
def loss(
self,
model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch,
) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss function.
Args:
model: The Model to calculate the loss for.
dist_class: The action distr. class.
train_batch: The training data.
Returns:
The A3C loss tensor given the input batch.
"""
logits, _ = model(train_batch)
values = model.value_function()
if self.is_recurrent():
B = len(train_batch[SampleBatch.SEQ_LENS])
max_seq_len = logits.shape[0] // B
mask_orig = sequence_mask(train_batch[SampleBatch.SEQ_LENS],
max_seq_len)
valid_mask = torch.reshape(mask_orig, [-1])
else:
valid_mask = torch.ones_like(values, dtype=torch.bool)
dist = dist_class(logits, model)
log_probs = dist.logp(train_batch[SampleBatch.ACTIONS]).reshape(-1)
pi_err = -torch.sum(
torch.masked_select(
log_probs * train_batch[Postprocessing.ADVANTAGES],
valid_mask))
# Compute a value function loss.
if self.config["use_critic"]:
value_err = 0.5 * torch.sum(
torch.pow(
torch.masked_select(
values.reshape(-1) -
train_batch[Postprocessing.VALUE_TARGETS],
valid_mask,
),
2.0,
))
# Ignore the value function.
else:
value_err = 0.0
entropy = torch.sum(torch.masked_select(dist.entropy(), valid_mask))
total_loss = (pi_err + value_err * self.config["vf_loss_coeff"] -
entropy * self.entropy_coeff)
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["entropy"] = entropy
model.tower_stats["pi_err"] = pi_err
model.tower_stats["value_err"] = value_err
return total_loss
def loss(
self,
model: ModelV2,
dist_class: Type[ActionDistribution],
train_batch: SampleBatch,
) -> Union[TensorType, List[TensorType]]:
model_out, _ = model(train_batch)
action_dist = dist_class(model_out, model)
if isinstance(self.action_space, gym.spaces.Discrete):
is_multidiscrete = False
output_hidden_shape = [self.action_space.n]
elif isinstance(self.action_space, gym.spaces.MultiDiscrete):
is_multidiscrete = True
output_hidden_shape = self.action_space.nvec.astype(np.int32)
else:
is_multidiscrete = False
output_hidden_shape = 1
def _make_time_major(*args, **kw):
return make_time_major(self, train_batch.get(SampleBatch.SEQ_LENS),
*args, **kw)
actions = train_batch[SampleBatch.ACTIONS]
dones = train_batch[SampleBatch.DONES]
rewards = train_batch[SampleBatch.REWARDS]
behaviour_action_logp = train_batch[SampleBatch.ACTION_LOGP]
behaviour_logits = train_batch[SampleBatch.ACTION_DIST_INPUTS]
if isinstance(output_hidden_shape, (list, tuple, np.ndarray)):
unpacked_behaviour_logits = torch.split(behaviour_logits,
list(output_hidden_shape),
dim=1)
unpacked_outputs = torch.split(model_out,
list(output_hidden_shape),
dim=1)
else:
unpacked_behaviour_logits = torch.chunk(behaviour_logits,
output_hidden_shape,
dim=1)
unpacked_outputs = torch.chunk(model_out,
output_hidden_shape,
dim=1)
values = model.value_function()
if self.is_recurrent():
max_seq_len = torch.max(train_batch[SampleBatch.SEQ_LENS])
mask_orig = sequence_mask(train_batch[SampleBatch.SEQ_LENS],
max_seq_len)
mask = torch.reshape(mask_orig, [-1])
else:
mask = torch.ones_like(rewards)
# Prepare actions for loss.
loss_actions = actions if is_multidiscrete else torch.unsqueeze(
actions, dim=1)
# Inputs are reshaped from [B * T] => [(T|T-1), B] for V-trace calc.
drop_last = self.config["vtrace_drop_last_ts"]
loss = VTraceLoss(
actions=_make_time_major(loss_actions, drop_last=drop_last),
actions_logp=_make_time_major(action_dist.logp(actions),
drop_last=drop_last),
actions_entropy=_make_time_major(action_dist.entropy(),
drop_last=drop_last),
dones=_make_time_major(dones, drop_last=drop_last),
behaviour_action_logp=_make_time_major(behaviour_action_logp,
drop_last=drop_last),
behaviour_logits=_make_time_major(unpacked_behaviour_logits,
drop_last=drop_last),
target_logits=_make_time_major(unpacked_outputs,
drop_last=drop_last),
discount=self.config["gamma"],
rewards=_make_time_major(rewards, drop_last=drop_last),
values=_make_time_major(values, drop_last=drop_last),
bootstrap_value=_make_time_major(values)[-1],
dist_class=TorchCategorical if is_multidiscrete else dist_class,
model=model,
valid_mask=_make_time_major(mask, drop_last=drop_last),
config=self.config,
vf_loss_coeff=self.config["vf_loss_coeff"],
entropy_coeff=self.entropy_coeff,
clip_rho_threshold=self.config["vtrace_clip_rho_threshold"],
clip_pg_rho_threshold=self.config["vtrace_clip_pg_rho_threshold"],
)
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["pi_loss"] = loss.pi_loss
model.tower_stats["vf_loss"] = loss.vf_loss
model.tower_stats["entropy"] = loss.entropy
model.tower_stats["mean_entropy"] = loss.mean_entropy
model.tower_stats["total_loss"] = loss.total_loss
values_batched = make_time_major(
self,
train_batch.get(SampleBatch.SEQ_LENS),
values,
drop_last=self.config["vtrace"] and drop_last,
)
model.tower_stats["vf_explained_var"] = explained_variance(
torch.reshape(loss.value_targets, [-1]),
torch.reshape(values_batched, [-1]))
return loss.total_loss
def build_slateq_losses(
policy: Policy,
model: ModelV2,
_: Type[TorchDistributionWrapper],
train_batch: SampleBatch,
) -> TensorType:
"""Constructs the choice- and Q-value losses for the SlateQTorchPolicy.
Args:
policy: The Policy to calculate the loss for.
model: The Model to calculate the loss for.
train_batch: The training data.
Returns:
Tuple consisting of 1) the choice loss- and 2) the Q-value loss tensors.
"""
start = time.time()
obs = restore_original_dimensions(train_batch[SampleBatch.OBS],
policy.observation_space,
tensorlib=torch)
# user.shape: [batch_size, embedding_size]
user = obs["user"]
# doc.shape: [batch_size, num_docs, embedding_size]
doc = torch.cat([val.unsqueeze(1) for val in obs["doc"].values()], 1)
# action.shape: [batch_size, slate_size]
actions = train_batch[SampleBatch.ACTIONS]
next_obs = restore_original_dimensions(train_batch[SampleBatch.NEXT_OBS],
policy.observation_space,
tensorlib=torch)
# Step 1: Build user choice model loss
_, _, embedding_size = doc.shape
# selected_doc.shape: [batch_size, slate_size, embedding_size]
selected_doc = torch.gather(
# input.shape: [batch_size, num_docs, embedding_size]
input=doc,
dim=1,
# index.shape: [batch_size, slate_size, embedding_size]
index=actions.unsqueeze(2).expand(-1, -1, embedding_size).long(),
)
scores = model.choice_model(user, selected_doc)
choice_loss_fn = nn.CrossEntropyLoss()
# clicks.shape: [batch_size, slate_size]
clicks = torch.stack(
[resp["click"][:, 1] for resp in next_obs["response"]], dim=1)
no_clicks = 1 - torch.sum(clicks, 1, keepdim=True)
# clicks.shape: [batch_size, slate_size+1]
targets = torch.cat([clicks, no_clicks], dim=1)
choice_loss = choice_loss_fn(scores, torch.argmax(targets, dim=1))
# print(model.choice_model.a.item(), model.choice_model.b.item())
# Step 2: Build qvalue loss
# Fields in available in train_batch: ['t', 'eps_id', 'agent_index',
# 'next_actions', 'obs', 'actions', 'rewards', 'prev_actions',
# 'prev_rewards', 'dones', 'infos', 'new_obs', 'unroll_id', 'weights',
# 'batch_indexes']
learning_strategy = policy.config["slateq_strategy"]
# Myopic agent: Don't care about value of next state.
# Acts only based off immediate reward.
if learning_strategy == "MYOP":
next_q_values = torch.tensor(0.0, requires_grad=False)
# Q-learning: Default setting for SlateQ -> Use DQN-style loss function.
elif learning_strategy == "QL":
# next_doc.shape: [batch_size, num_docs, embedding_size]
next_doc = torch.cat(
[val.unsqueeze(1) for val in next_obs["doc"].values()], 1)
next_user = next_obs["user"]
dones = train_batch[SampleBatch.DONES]
with torch.no_grad():
if policy.config["double_q"]:
next_target_per_slate_q_values = policy.target_models[
model].get_per_slate_q_values(next_user, next_doc)
_, next_q_values, _ = model.choose_slate(
next_user, next_doc, next_target_per_slate_q_values)
else:
_, next_q_values, _ = policy.target_models[model].choose_slate(
next_user, next_doc)
next_q_values = next_q_values.detach()
next_q_values[dones.bool()] = 0.0
# SARS'A': Use on-policy sarsa loss.
elif learning_strategy == "SARSA":
# next_doc.shape: [batch_size, num_docs, embedding_size]
next_doc = torch.cat(
[val.unsqueeze(1) for val in next_obs["doc"].values()], 1)
next_actions = train_batch["next_actions"]
_, _, embedding_size = next_doc.shape
# selected_doc.shape: [batch_size, slate_size, embedding_size]
next_selected_doc = torch.gather(
# input.shape: [batch_size, num_docs, embedding_size]
input=next_doc,
dim=1,
# index.shape: [batch_size, slate_size, embedding_size]
index=next_actions.unsqueeze(2).expand(-1, -1,
embedding_size).long(),
)
next_user = next_obs["user"]
dones = train_batch[SampleBatch.DONES]
with torch.no_grad():
# q_values.shape: [batch_size, slate_size+1]
q_values = model.q_model(next_user, next_selected_doc)
# raw_scores.shape: [batch_size, slate_size+1]
raw_scores = model.choice_model(next_user, next_selected_doc)
max_raw_scores, _ = torch.max(raw_scores, dim=1, keepdim=True)
scores = torch.exp(raw_scores - max_raw_scores)
# next_q_values.shape: [batch_size]
next_q_values = torch.sum(q_values * scores, dim=1) / torch.sum(
scores, dim=1)
next_q_values[dones.bool()] = 0.0
else:
raise ValueError(learning_strategy)
# target_q_values.shape: [batch_size]
target_q_values = (train_batch[SampleBatch.REWARDS] +
policy.config["gamma"] * next_q_values)
# q_values.shape: [batch_size, slate_size+1].
q_values = model.q_model(user, selected_doc)
# raw_scores.shape: [batch_size, slate_size+1].
raw_scores = model.choice_model(user, selected_doc)
max_raw_scores, _ = torch.max(raw_scores, dim=1, keepdim=True)
scores = torch.exp(raw_scores - max_raw_scores)
q_values = torch.sum(q_values * scores, dim=1) / torch.sum(
scores, dim=1) # shape=[batch_size]
td_error = torch.abs(q_values - target_q_values)
q_value_loss = torch.mean(huber_loss(td_error))
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["q_loss"] = q_value_loss
model.tower_stats["q_values"] = q_values
model.tower_stats["next_q_values"] = next_q_values
model.tower_stats["next_q_minus_q"] = next_q_values - q_values
model.tower_stats["td_error"] = td_error
model.tower_stats["target_q_values"] = target_q_values
model.tower_stats["scores"] = scores
model.tower_stats["raw_scores"] = raw_scores
model.tower_stats["choice_loss"] = choice_loss
model.tower_stats["choice_beta"] = model.choice_model.beta
model.tower_stats[
"choice_score_no_click"] = model.choice_model.score_no_click
logger.debug(f"loss calculation took {time.time()-start}s")
return choice_loss, q_value_loss
def build_slateq_losses(
policy: Policy,
model: ModelV2,
_,
train_batch: SampleBatch,
) -> TensorType:
"""Constructs the choice- and Q-value losses for the SlateQTorchPolicy.
Args:
policy: The Policy to calculate the loss for.
model: The Model to calculate the loss for.
train_batch: The training data.
Returns:
The user-choice- and Q-value loss tensors.
"""
# B=batch size
# S=slate size
# C=num candidates
# E=embedding size
# A=number of all possible slates
# Q-value computations.
# ---------------------
# action.shape: [B, S]
actions = train_batch[SampleBatch.ACTIONS]
observation = convert_to_torch_tensor(
train_batch[SampleBatch.OBS], device=actions.device
)
# user.shape: [B, E]
user_obs = observation["user"]
batch_size, embedding_size = user_obs.shape
# doc.shape: [B, C, E]
doc_obs = list(observation["doc"].values())
A, S = policy.slates.shape
# click_indicator.shape: [B, S]
click_indicator = torch.stack(
[k["click"] for k in observation["response"]], 1
).float()
# item_reward.shape: [B, S]
item_reward = torch.stack([k["watch_time"] for k in observation["response"]], 1)
# q_values.shape: [B, C]
q_values = model.get_q_values(user_obs, doc_obs)
# slate_q_values.shape: [B, S]
slate_q_values = torch.take_along_dim(q_values, actions.long(), dim=-1)
# Only get the Q from the clicked document.
# replay_click_q.shape: [B]
replay_click_q = torch.sum(slate_q_values * click_indicator, dim=1)
# Target computations.
# --------------------
next_obs = convert_to_torch_tensor(
train_batch[SampleBatch.NEXT_OBS], device=actions.device
)
# user.shape: [B, E]
user_next_obs = next_obs["user"]
# doc.shape: [B, C, E]
doc_next_obs = list(next_obs["doc"].values())
# Only compute the watch time reward of the clicked item.
reward = torch.sum(item_reward * click_indicator, dim=1)
# TODO: Find out, whether it's correct here to use obs, not next_obs!
# Dopamine uses obs, then next_obs only for the score.
# next_q_values = policy.target_model.get_q_values(user_next_obs, doc_next_obs)
next_q_values = policy.target_models[model].get_q_values(user_obs, doc_obs)
scores, score_no_click = score_documents(user_next_obs, doc_next_obs)
# next_q_values_slate.shape: [B, A, S]
indices = policy.slates_indices.to(next_q_values.device)
next_q_values_slate = torch.take_along_dim(next_q_values, indices, dim=1).reshape(
[-1, A, S]
)
# scores_slate.shape [B, A, S]
scores_slate = torch.take_along_dim(scores, indices, dim=1).reshape([-1, A, S])
# score_no_click_slate.shape: [B, A]
score_no_click_slate = torch.reshape(
torch.tile(score_no_click, policy.slates.shape[:1]), [batch_size, -1]
)
# next_q_target_slate.shape: [B, A]
next_q_target_slate = torch.sum(next_q_values_slate * scores_slate, dim=2) / (
torch.sum(scores_slate, dim=2) + score_no_click_slate
)
next_q_target_max, _ = torch.max(next_q_target_slate, dim=1)
target = reward + policy.config["gamma"] * next_q_target_max * (
1.0 - train_batch["dones"].float()
)
target = target.detach()
clicked = torch.sum(click_indicator, dim=1)
mask_clicked_slates = clicked > 0
clicked_indices = torch.arange(batch_size).to(mask_clicked_slates.device)
clicked_indices = torch.masked_select(clicked_indices, mask_clicked_slates)
# Clicked_indices is a vector and torch.gather selects the batch dimension.
q_clicked = torch.gather(replay_click_q, 0, clicked_indices)
target_clicked = torch.gather(target, 0, clicked_indices)
td_error = torch.where(
clicked.bool(),
replay_click_q - target,
torch.zeros_like(train_batch[SampleBatch.REWARDS]),
)
if policy.config["use_huber"]:
loss = huber_loss(td_error, delta=policy.config["huber_threshold"])
else:
loss = torch.pow(td_error, 2.0)
loss = torch.mean(loss)
td_error = torch.abs(td_error)
mean_td_error = torch.mean(td_error)
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["q_values"] = torch.mean(q_values)
model.tower_stats["q_clicked"] = torch.mean(q_clicked)
model.tower_stats["scores"] = torch.mean(scores)
model.tower_stats["score_no_click"] = torch.mean(score_no_click)
model.tower_stats["slate_q_values"] = torch.mean(slate_q_values)
model.tower_stats["replay_click_q"] = torch.mean(replay_click_q)
model.tower_stats["bellman_reward"] = torch.mean(reward)
model.tower_stats["next_q_values"] = torch.mean(next_q_values)
model.tower_stats["target"] = torch.mean(target)
model.tower_stats["next_q_target_slate"] = torch.mean(next_q_target_slate)
model.tower_stats["next_q_target_max"] = torch.mean(next_q_target_max)
model.tower_stats["target_clicked"] = torch.mean(target_clicked)
model.tower_stats["q_loss"] = loss
model.tower_stats["td_error"] = td_error
model.tower_stats["mean_td_error"] = mean_td_error
model.tower_stats["mean_actions"] = torch.mean(actions.float())
# selected_doc.shape: [batch_size, slate_size, embedding_size]
selected_doc = torch.gather(
# input.shape: [batch_size, num_docs, embedding_size]
torch.stack(doc_obs, 1),
1,
# index.shape: [batch_size, slate_size, embedding_size]
actions.unsqueeze(2).expand(-1, -1, embedding_size).long(),
)
scores = model.choice_model(user_obs, selected_doc)
# click_indicator.shape: [batch_size, slate_size]
# no_clicks.shape: [batch_size, 1]
no_clicks = 1 - torch.sum(click_indicator, 1, keepdim=True)
# targets.shape: [batch_size, slate_size+1]
targets = torch.cat([click_indicator, no_clicks], dim=1)
choice_loss = nn.functional.cross_entropy(scores, torch.argmax(targets, dim=1))
# print(model.choice_model.a.item(), model.choice_model.b.item())
model.tower_stats["choice_loss"] = choice_loss
return choice_loss, loss
def _compute_critic_loss(
self,
model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch,
):
discount = self.config["gamma"]
# Compute bellman targets to regress on
# target, use target model to compute the target
target_model = cast(CRRModel, self.target_models[model])
target_out_next, _ = target_model(
{SampleBatch.OBS: train_batch[SampleBatch.NEXT_OBS]}
)
# compute target values with no gradient
with torch.no_grad():
# get the action of the current policy evaluated at the next state
pi_s_next = dist_class(
target_model.get_policy_output(target_out_next), target_model
)
target_a_next = pi_s_next.sample()
if not self._is_action_discrete:
target_a_next = target_a_next.clamp(
torch.from_numpy(self.action_space.low).to(target_a_next),
torch.from_numpy(self.action_space.high).to(target_a_next),
)
# q1_target = target_model.get_q_values(target_out_next, target_a_next)
# q2_target = target_model.get_twin_q_values(target_out_next, target_a_next)
# target_q_next = torch.minimum(q1_target, q2_target).squeeze(-1)
target_q_next = self._get_q_value(
target_model, target_out_next, target_a_next
).squeeze(-1)
target = (
train_batch[SampleBatch.REWARDS]
+ discount
* (1.0 - train_batch[SampleBatch.DONES].float())
* target_q_next
)
# compute the predicted output
model = cast(CRRModel, model)
model_out_t, _ = model({SampleBatch.OBS: train_batch[SampleBatch.OBS]})
q1 = model.get_q_values(model_out_t, train_batch[SampleBatch.ACTIONS]).squeeze(
-1
)
q2 = model.get_twin_q_values(
model_out_t, train_batch[SampleBatch.ACTIONS]
).squeeze(-1)
# compute the MSE loss for all q-functions
loss_q1 = (target - q1) ** 2
loss_q2 = (target - q2) ** 2
loss = 0.5 * (loss_q1 + loss_q2)
loss = loss.mean(0)
# logging
self.log("loss_q1", loss_q1.mean())
self.log("loss_q2", loss_q2.mean())
self.log("targets_avg", target.mean())
self.log("targets_max", target.max())
self.log("targets_min", target.min())
return loss
def _compute_adv_and_logps(
self,
model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch,
) -> None:
# uses mean|max|expectation to compute estimate of advantages
# continuous/discrete action spaces:
# for max:
# A(s_t, a_t) = Q(s_t, a_t) - max_{a^j} Q(s_t, a^j)
# where a^j is m times sampled from the policy p(a | s_t)
# for mean:
# A(s_t, a_t) = Q(s_t, a_t) - avg( Q(s_t, a^j) )
# where a^j is m times sampled from the policy p(a | s_t)
# discrete action space and adv_type=expectation:
# A(s_t, a_t) = Q(s_t, a_t) - sum_j[Q(s_t, a^j) * pi(a^j)]
advantage_type = self.config["advantage_type"]
n_action_sample = self.config["n_action_sample"]
batch_size = len(train_batch)
out_t, _ = model(train_batch)
# construct pi(s_t) and Q(s_t, a_t) for computing advantage actions
pi_s_t = dist_class(model.get_policy_output(out_t), model)
q_t = self._get_q_value(model, out_t, train_batch[SampleBatch.ACTIONS])
# compute the logp of the actions in the dataset (for computing actor's loss)
action_logp = pi_s_t.dist.log_prob(train_batch[SampleBatch.ACTIONS])
# fix the shape if it's not canonical (i.e. shape[-1] != 1)
if len(action_logp.shape) <= 1:
action_logp.unsqueeze_(-1)
train_batch[SampleBatch.ACTION_LOGP] = action_logp
if advantage_type == "expectation":
assert (
self._is_action_discrete
), "Action space should be discrete when advantage_type = expectation."
assert hasattr(
self.model, "q_model"
), "CRR's ModelV2 should have q_model neural network in discrete \
action spaces"
assert isinstance(
pi_s_t.dist, torch.distributions.Categorical
), "The output of the policy should be a torch Categorical \
distribution."
q_vals = self.model.q_model(out_t)
if hasattr(self.model, "twin_q_model"):
q_twins = self.model.twin_q_model(out_t)
q_vals = torch.minimum(q_vals, q_twins)
probs = pi_s_t.dist.probs
v_t = (q_vals * probs).sum(-1, keepdims=True)
else:
policy_actions = pi_s_t.dist.sample((n_action_sample,)) # samples
if self._is_action_discrete:
flat_actions = policy_actions.reshape(-1)
else:
flat_actions = policy_actions.reshape(-1, *self.action_space.shape)
reshaped_s_t = train_batch[SampleBatch.OBS].view(
1, batch_size, *self.observation_space.shape
)
reshaped_s_t = reshaped_s_t.expand(
n_action_sample, batch_size, *self.observation_space.shape
)
flat_s_t = reshaped_s_t.reshape(-1, *self.observation_space.shape)
input_v_t = SampleBatch(
**{SampleBatch.OBS: flat_s_t, SampleBatch.ACTIONS: flat_actions}
)
out_v_t, _ = model(input_v_t)
flat_q_st_pi = self._get_q_value(model, out_v_t, flat_actions)
reshaped_q_st_pi = flat_q_st_pi.reshape(-1, batch_size, 1)
if advantage_type == "mean":
v_t = reshaped_q_st_pi.mean(dim=0)
elif advantage_type == "max":
v_t, _ = reshaped_q_st_pi.max(dim=0)
else:
raise ValueError(f"Invalid advantage type: {advantage_type}.")
adv_t = q_t - v_t
train_batch["advantages"] = adv_t
# logging
self.log("q_batch_avg", q_t.mean())
self.log("q_batch_max", q_t.max())
self.log("q_batch_min", q_t.min())
self.log("v_batch_avg", v_t.mean())
self.log("v_batch_max", v_t.max())
self.log("v_batch_min", v_t.min())
self.log("adv_batch_avg", adv_t.mean())
self.log("adv_batch_max", adv_t.max())
self.log("adv_batch_min", adv_t.min())
self.log("reward_batch_avg", train_batch[SampleBatch.REWARDS].mean())
def build_slateq_losses(
policy: Policy,
model: ModelV2,
_,
train_batch: SampleBatch,
) -> TensorType:
"""Constructs the choice- and Q-value losses for the SlateQTorchPolicy.
Args:
policy: The Policy to calculate the loss for.
model: The Model to calculate the loss for.
train_batch: The training data.
Returns:
The Q-value loss tensor.
"""
# B=batch size
# S=slate size
# C=num candidates
# E=embedding size
# A=number of all possible slates
# Q-value computations.
# ---------------------
observation = train_batch[SampleBatch.OBS]
# user.shape: [B, E]
user_obs = observation["user"]
batch_size = tf.shape(user_obs)[0]
# doc.shape: [B, C, E]
doc_obs = list(observation["doc"].values())
# action.shape: [B, S]
actions = train_batch[SampleBatch.ACTIONS]
# click_indicator.shape: [B, S]
click_indicator = tf.cast(
tf.stack([k["click"] for k in observation["response"]], 1), tf.float32)
# item_reward.shape: [B, S]
item_reward = tf.stack([k["watch_time"] for k in observation["response"]],
1)
# q_values.shape: [B, C]
q_values = model.get_q_values(user_obs, doc_obs)
# slate_q_values.shape: [B, S]
slate_q_values = tf.gather(q_values,
tf.cast(actions, dtype=tf.int32),
batch_dims=-1)
# Only get the Q from the clicked document.
# replay_click_q.shape: [B]
replay_click_q = tf.reduce_sum(input_tensor=slate_q_values *
click_indicator,
axis=1,
name="replay_click_q")
# Target computations.
# --------------------
next_obs = train_batch[SampleBatch.NEXT_OBS]
# user.shape: [B, E]
user_next_obs = next_obs["user"]
# doc.shape: [B, C, E]
doc_next_obs = list(next_obs["doc"].values())
# Only compute the watch time reward of the clicked item.
reward = tf.reduce_sum(input_tensor=item_reward * click_indicator, axis=1)
# TODO: Find out, whether it's correct here to use obs, not next_obs!
# Dopamine uses obs, then next_obs only for the score.
# next_q_values = policy.target_model.get_q_values(user_next_obs, doc_next_obs)
next_q_values = policy.target_model.get_q_values(user_obs, doc_obs)
scores, score_no_click = score_documents(user_next_obs, doc_next_obs)
# next_q_values_slate.shape: [B, A, S]
next_q_values_slate = tf.gather(next_q_values, policy.slates, axis=1)
# scores_slate.shape [B, A, S]
scores_slate = tf.gather(scores, policy.slates, axis=1)
# score_no_click_slate.shape: [B, A]
score_no_click_slate = tf.reshape(
tf.tile(score_no_click,
tf.shape(input=policy.slates)[:1]), [batch_size, -1])
# next_q_target_slate.shape: [B, A]
next_q_target_slate = tf.reduce_sum(
input_tensor=next_q_values_slate * scores_slate,
axis=2) / (tf.reduce_sum(input_tensor=scores_slate, axis=2) +
score_no_click_slate)
next_q_target_max = tf.reduce_max(input_tensor=next_q_target_slate, axis=1)
target = reward + policy.config["gamma"] * next_q_target_max * (
1.0 - tf.cast(train_batch["dones"], tf.float32))
target = tf.stop_gradient(target)
clicked = tf.reduce_sum(input_tensor=click_indicator, axis=1)
clicked_indices = tf.squeeze(tf.where(tf.equal(clicked, 1)), axis=1)
# Clicked_indices is a vector and tf.gather selects the batch dimension.
q_clicked = tf.gather(replay_click_q, clicked_indices)
target_clicked = tf.gather(target, clicked_indices)
td_error = tf.where(
tf.cast(clicked, tf.bool),
replay_click_q - target,
tf.zeros_like(train_batch[SampleBatch.REWARDS]),
)
if policy.config["use_huber"]:
loss = huber_loss(td_error, delta=policy.config["huber_threshold"])
else:
loss = tf.math.square(td_error)
loss = tf.reduce_mean(loss)
td_error = tf.abs(td_error)
mean_td_error = tf.reduce_mean(td_error)
policy._q_values = tf.reduce_mean(q_values)
policy._q_clicked = tf.reduce_mean(q_clicked)
policy._scores = tf.reduce_mean(scores)
policy._score_no_click = tf.reduce_mean(score_no_click)
policy._slate_q_values = tf.reduce_mean(slate_q_values)
policy._replay_click_q = tf.reduce_mean(replay_click_q)
policy._bellman_reward = tf.reduce_mean(reward)
policy._next_q_values = tf.reduce_mean(next_q_values)
policy._target = tf.reduce_mean(target)
policy._next_q_target_slate = tf.reduce_mean(next_q_target_slate)
policy._next_q_target_max = tf.reduce_mean(next_q_target_max)
policy._target_clicked = tf.reduce_mean(target_clicked)
policy._q_loss = loss
policy._td_error = td_error
policy._mean_td_error = mean_td_error
policy._mean_actions = tf.reduce_mean(actions)
return loss
def __init__(
self,
observation_space: gym.spaces.Space,
action_space: gym.spaces.Space,
config: TrainerConfigDict,
*,
model: ModelV2,
loss: Callable[
[Policy, ModelV2, Type[TorchDistributionWrapper], SampleBatch],
Union[TensorType, List[TensorType]]],
action_distribution_class: Type[TorchDistributionWrapper],
action_sampler_fn: Optional[Callable[[TensorType, List[TensorType]],
Tuple[TensorType,
TensorType]]] = None,
action_distribution_fn: Optional[
Callable[[Policy, ModelV2, TensorType, TensorType, TensorType],
Tuple[TensorType, Type[TorchDistributionWrapper],
List[TensorType]]]] = None,
max_seq_len: int = 20,
get_batch_divisibility_req: Optional[Callable[[Policy], int]] = None,
):
"""Build a policy from policy and loss torch modules.
Note that model will be placed on GPU device if CUDA_VISIBLE_DEVICES
is set. Only single GPU is supported for now.
Args:
observation_space (gym.spaces.Space): observation space of the
policy.
action_space (gym.spaces.Space): action space of the policy.
config (TrainerConfigDict): The Policy config dict.
model (ModelV2): PyTorch policy module. Given observations as
input, this module must return a list of outputs where the
first item is action logits, and the rest can be any value.
loss (Callable[[Policy, ModelV2, Type[TorchDistributionWrapper],
SampleBatch], Union[TensorType, List[TensorType]]]): Callable
that returns a single scalar loss or a list of loss terms.
action_distribution_class (Type[TorchDistributionWrapper]): Class
for a torch action distribution.
action_sampler_fn (Callable[[TensorType, List[TensorType]],
Tuple[TensorType, TensorType]]): A callable returning a
sampled action and its log-likelihood given Policy, ModelV2,
input_dict, explore, timestep, and is_training.
action_distribution_fn (Optional[Callable[[Policy, ModelV2,
Dict[str, TensorType], TensorType, TensorType],
Tuple[TensorType, type, List[TensorType]]]]): A callable
returning distribution inputs (parameters), a dist-class to
generate an action distribution object from, and
internal-state outputs (or an empty list if not applicable).
Note: No Exploration hooks have to be called from within
`action_distribution_fn`. It's should only perform a simple
forward pass through some model.
If None, pass inputs through `self.model()` to get distribution
inputs.
The callable takes as inputs: Policy, ModelV2, input_dict,
explore, timestep, is_training.
max_seq_len (int): Max sequence length for LSTM training.
get_batch_divisibility_req (Optional[Callable[[Policy], int]]]):
Optional callable that returns the divisibility requirement
for sample batches given the Policy.
"""
self.framework = "torch"
super().__init__(observation_space, action_space, config)
if torch.cuda.is_available():
logger.info("TorchPolicy running on GPU.")
self.device = torch.device("cuda")
else:
logger.info("TorchPolicy running on CPU.")
self.device = torch.device("cpu")
self.model = model.to(self.device)
# Combine view_requirements for Model and Policy.
self.view_requirements.update(self.model.inference_view_requirements)
self.exploration = self._create_exploration()
self.unwrapped_model = model # used to support DistributedDataParallel
self._loss = loss
self._optimizers = force_list(self.optimizer())
self.dist_class = action_distribution_class
self.action_sampler_fn = action_sampler_fn
self.action_distribution_fn = action_distribution_fn
# If set, means we are using distributed allreduce during learning.
self.distributed_world_size = None
self.max_seq_len = max_seq_len
self.batch_divisibility_req = get_batch_divisibility_req(self) if \
callable(get_batch_divisibility_req) else \
(get_batch_divisibility_req or 1)
def action_distribution_fn(
policy: Policy,
model: ModelV2,
input_dict: ModelInputDict,
*,
state_batches: Optional[List[TensorType]] = None,
seq_lens: Optional[TensorType] = None,
prev_action_batch: Optional[TensorType] = None,
prev_reward_batch=None,
explore: Optional[bool] = None,
timestep: Optional[int] = None,
is_training: Optional[bool] = None) -> \
Tuple[TensorType, Type[TorchDistributionWrapper], List[TensorType]]:
"""The action distribution function to be used the algorithm.
An action distribution function is used to customize the choice of action
distribution class and the resulting action distribution inputs (to
parameterize the distribution object).
After parameterizing the distribution, a `sample()` call
will be made on it to generate actions.
Args:
policy (Policy): The Policy being queried for actions and calling this
function.
model (TorchModelV2): The SAC specific Model to use to generate the
distribution inputs (see sac_tf|torch_model.py). Must support the
`get_policy_output` method.
input_dict (ModelInputDict): The input-dict to be used for the model
call.
state_batches (Optional[List[TensorType]]): The list of internal state
tensor batches.
seq_lens (Optional[TensorType]): The tensor of sequence lengths used
in RNNs.
prev_action_batch (Optional[TensorType]): Optional batch of prev
actions used by the model.
prev_reward_batch (Optional[TensorType]): Optional batch of prev
rewards used by the model.
explore (Optional[bool]): Whether to activate exploration or not. If
None, use value of `config.explore`.
timestep (Optional[int]): An optional timestep.
is_training (Optional[bool]): An optional is-training flag.
Returns:
Tuple[TensorType, Type[TorchDistributionWrapper], List[TensorType]]:
The dist inputs, dist class, and a list of internal state outputs
(in the RNN case).
"""
# Get base-model output (w/o the SAC specific parts of the network).
model_out, state_in = model(input_dict, state_batches, seq_lens)
# Use the base output to get the policy outputs from the SAC model's
# policy components.
states_in = model.select_state(state_in, ["policy", "q", "twin_q"])
distribution_inputs, policy_state_out = \
model.get_policy_output(model_out, states_in["policy"], seq_lens)
_, q_state_out = model.get_q_values(model_out, states_in["q"], seq_lens)
if model.twin_q_net:
_, twin_q_state_out = \
model.get_twin_q_values(model_out, states_in["twin_q"], seq_lens)
else:
twin_q_state_out = []
# Get a distribution class to be used with the just calculated dist-inputs.
action_dist_class = _get_dist_class(policy, policy.config,
policy.action_space)
states_out = policy_state_out + q_state_out + twin_q_state_out
return distribution_inputs, action_dist_class, states_out
def __init__(
self,
observation_space: gym.spaces.Space,
action_space: gym.spaces.Space,
config: TrainerConfigDict,
*,
model: ModelV2,
loss: Callable[[
Policy, ModelV2, Type[TorchDistributionWrapper], SampleBatch
], Union[TensorType, List[TensorType]]],
action_distribution_class: Type[TorchDistributionWrapper],
action_sampler_fn: Optional[Callable[[
TensorType, List[TensorType]
], Tuple[TensorType, TensorType]]] = None,
action_distribution_fn: Optional[Callable[[
Policy, ModelV2, TensorType, TensorType, TensorType
], Tuple[TensorType, Type[TorchDistributionWrapper], List[
TensorType]]]] = None,
max_seq_len: int = 20,
get_batch_divisibility_req: Optional[Callable[[Policy],
int]] = None,
):
"""Build a policy from policy and loss torch modules.
Note that model will be placed on GPU device if CUDA_VISIBLE_DEVICES
is set. Only single GPU is supported for now.
Args:
observation_space (gym.spaces.Space): observation space of the
policy.
action_space (gym.spaces.Space): action space of the policy.
config (TrainerConfigDict): The Policy config dict.
model (ModelV2): PyTorch policy module. Given observations as
input, this module must return a list of outputs where the
first item is action logits, and the rest can be any value.
loss (Callable[[Policy, ModelV2, Type[TorchDistributionWrapper],
SampleBatch], Union[TensorType, List[TensorType]]]): Callable
that returns a single scalar loss or a list of loss terms.
action_distribution_class (Type[TorchDistributionWrapper]): Class
for a torch action distribution.
action_sampler_fn (Callable[[TensorType, List[TensorType]],
Tuple[TensorType, TensorType]]): A callable returning a
sampled action and its log-likelihood given Policy, ModelV2,
input_dict, explore, timestep, and is_training.
action_distribution_fn (Optional[Callable[[Policy, ModelV2,
ModelInputDict, TensorType, TensorType],
Tuple[TensorType, type, List[TensorType]]]]): A callable
returning distribution inputs (parameters), a dist-class to
generate an action distribution object from, and
internal-state outputs (or an empty list if not applicable).
Note: No Exploration hooks have to be called from within
`action_distribution_fn`. It's should only perform a simple
forward pass through some model.
If None, pass inputs through `self.model()` to get distribution
inputs.
The callable takes as inputs: Policy, ModelV2, ModelInputDict,
explore, timestep, is_training.
max_seq_len (int): Max sequence length for LSTM training.
get_batch_divisibility_req (Optional[Callable[[Policy], int]]]):
Optional callable that returns the divisibility requirement
for sample batches given the Policy.
"""
self.framework = "torch"
super().__init__(observation_space, action_space, config)
# Log device and worker index.
from ray.rllib.evaluation.rollout_worker import get_global_worker
worker = get_global_worker()
worker_idx = worker.worker_index if worker else 0
# Create multi-GPU model towers, if necessary.
# - The central main model will be stored under self.model, residing on
# self.device.
# - Each GPU will have a copy of that model under
# self.model_gpu_towers, matching the devices in self.devices.
# - Parallelization is done by splitting the train batch and passing
# it through the model copies in parallel, then averaging over the
# resulting gradients, applying these averages on the main model and
# updating all towers' weights from the main model.
# - In case of just one device (1 (fake) GPU or 1 CPU), no
# parallelization will be done.
if config["_fake_gpus"] or config["num_gpus"] == 0 or \
not torch.cuda.is_available():
logger.info("TorchPolicy (worker={}) running on {}.".format(
worker_idx if worker_idx > 0 else "local",
"{} fake-GPUs".format(config["num_gpus"])
if config["_fake_gpus"] else "CPU"))
self.device = torch.device("cpu")
self.devices = [
self.device for _ in range(config["num_gpus"] or 1)
]
self.model_gpu_towers = [
model if config["num_gpus"] == 0 else copy.deepcopy(model)
for i in range(config["num_gpus"] or 1)
]
else:
logger.info("TorchPolicy (worker={}) running on {} GPU(s).".format(
worker_idx if worker_idx > 0 else "local", config["num_gpus"]))
self.device = torch.device("cuda")
self.devices = [
torch.device("cuda:{}".format(id_))
for i, id_ in enumerate(ray.get_gpu_ids())
if i < config["num_gpus"]
]
self.model_gpu_towers = nn.parallel.replicate.replicate(
model, [
id_ for i, id_ in enumerate(ray.get_gpu_ids())
if i < config["num_gpus"]
])
# Move model to device.
self.model = model.to(self.device)
# Lock used for locking some methods on the object-level.
# This prevents possible race conditions when calling the model
# first, then its value function (e.g. in a loss function), in
# between of which another model call is made (e.g. to compute an
# action).
self._lock = threading.RLock()
self._state_inputs = self.model.get_initial_state()
self._is_recurrent = len(self._state_inputs) > 0
# Auto-update model's inference view requirements, if recurrent.
self._update_model_view_requirements_from_init_state()
# Combine view_requirements for Model and Policy.
self.view_requirements.update(self.model.view_requirements)
self.exploration = self._create_exploration()
self.unwrapped_model = model # used to support DistributedDataParallel
self._loss = loss
self._optimizers = force_list(self.optimizer())
# Store, which params (by index within the model's list of
# parameters) should be updated per optimizer.
# Maps optimizer idx to set or param indices.
self.multi_gpu_param_groups: List[Set[int]] = []
main_params = {p: i for i, p in enumerate(self.model.parameters())}
for o in self._optimizers:
param_indices = []
for pg_idx, pg in enumerate(o.param_groups):
for p in pg["params"]:
param_indices.append(main_params[p])
self.multi_gpu_param_groups.append(set(param_indices))
self.dist_class = action_distribution_class
self.action_sampler_fn = action_sampler_fn
self.action_distribution_fn = action_distribution_fn
# If set, means we are using distributed allreduce during learning.
self.distributed_world_size = None
self.max_seq_len = max_seq_len
self.batch_divisibility_req = get_batch_divisibility_req(self) if \
callable(get_batch_divisibility_req) else \
(get_batch_divisibility_req or 1)
def actor_critic_loss(
policy: Policy, model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss for the Soft Actor Critic.
Args:
policy (Policy): The Policy to calculate the loss for.
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[TorchDistributionWrapper]: The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
target_model = policy.target_models[model]
# Should be True only for debugging purposes (e.g. test cases)!
deterministic = policy.config["_deterministic_loss"]
i = 0
state_batches = []
while "state_in_{}".format(i) in train_batch:
state_batches.append(train_batch["state_in_{}".format(i)])
i += 1
assert state_batches
seq_lens = train_batch.get(SampleBatch.SEQ_LENS)
model_out_t, state_in_t = model(
{
"obs": train_batch[SampleBatch.CUR_OBS],
"prev_actions": train_batch[SampleBatch.PREV_ACTIONS],
"prev_rewards": train_batch[SampleBatch.PREV_REWARDS],
"is_training": True,
}, state_batches, seq_lens)
states_in_t = model.select_state(state_in_t, ["policy", "q", "twin_q"])
model_out_tp1, state_in_tp1 = model(
{
"obs": train_batch[SampleBatch.NEXT_OBS],
"prev_actions": train_batch[SampleBatch.ACTIONS],
"prev_rewards": train_batch[SampleBatch.REWARDS],
"is_training": True,
}, state_batches, seq_lens)
states_in_tp1 = model.select_state(state_in_tp1, ["policy", "q", "twin_q"])
target_model_out_tp1, target_state_in_tp1 = target_model(
{
"obs": train_batch[SampleBatch.NEXT_OBS],
"prev_actions": train_batch[SampleBatch.ACTIONS],
"prev_rewards": train_batch[SampleBatch.REWARDS],
"is_training": True,
}, state_batches, seq_lens)
target_states_in_tp1 = target_model.select_state(state_in_tp1,
["policy", "q", "twin_q"])
alpha = torch.exp(model.log_alpha)
# Discrete case.
if model.discrete:
# Get all action probs directly from pi and form their logp.
log_pis_t = F.log_softmax(model.get_policy_output(
model_out_t, states_in_t["policy"], seq_lens)[0],
dim=-1)
policy_t = torch.exp(log_pis_t)
log_pis_tp1 = F.log_softmax(
model.get_policy_output(model_out_tp1, states_in_tp1["policy"],
seq_lens)[0], -1)
policy_tp1 = torch.exp(log_pis_tp1)
# Q-values.
q_t = model.get_q_values(model_out_t, states_in_t["q"], seq_lens)[0]
# Target Q-values.
q_tp1 = target_model.get_q_values(target_model_out_tp1,
target_states_in_tp1["q"],
seq_lens)[0]
if policy.config["twin_q"]:
twin_q_t = model.get_twin_q_values(model_out_t,
states_in_t["twin_q"],
seq_lens)[0]
twin_q_tp1 = target_model.get_twin_q_values(
target_model_out_tp1, target_states_in_tp1["twin_q"],
seq_lens)[0]
q_tp1 = torch.min(q_tp1, twin_q_tp1)
q_tp1 -= alpha * log_pis_tp1
# Actually selected Q-values (from the actions batch).
one_hot = F.one_hot(train_batch[SampleBatch.ACTIONS].long(),
num_classes=q_t.size()[-1])
q_t_selected = torch.sum(q_t * one_hot, dim=-1)
if policy.config["twin_q"]:
twin_q_t_selected = torch.sum(twin_q_t * one_hot, dim=-1)
# Discrete case: "Best" means weighted by the policy (prob) outputs.
q_tp1_best = torch.sum(torch.mul(policy_tp1, q_tp1), dim=-1)
q_tp1_best_masked = \
(1.0 - train_batch[SampleBatch.DONES].float()) * \
q_tp1_best
# Continuous actions case.
else:
# Sample single actions from distribution.
action_dist_class = _get_dist_class(policy, policy.config,
policy.action_space)
action_dist_t = action_dist_class(
model.get_policy_output(model_out_t, states_in_t["policy"],
seq_lens)[0], model)
policy_t = action_dist_t.sample() if not deterministic else \
action_dist_t.deterministic_sample()
log_pis_t = torch.unsqueeze(action_dist_t.logp(policy_t), -1)
action_dist_tp1 = action_dist_class(
model.get_policy_output(model_out_tp1, states_in_tp1["policy"],
seq_lens)[0], model)
policy_tp1 = action_dist_tp1.sample() if not deterministic else \
action_dist_tp1.deterministic_sample()
log_pis_tp1 = torch.unsqueeze(action_dist_tp1.logp(policy_tp1), -1)
# Q-values for the actually selected actions.
q_t = model.get_q_values(model_out_t, states_in_t["q"], seq_lens,
train_batch[SampleBatch.ACTIONS])[0]
if policy.config["twin_q"]:
twin_q_t = model.get_twin_q_values(
model_out_t, states_in_t["twin_q"], seq_lens,
train_batch[SampleBatch.ACTIONS])[0]
# Q-values for current policy in given current state.
q_t_det_policy = model.get_q_values(model_out_t, states_in_t["q"],
seq_lens, policy_t)[0]
if policy.config["twin_q"]:
twin_q_t_det_policy = model.get_twin_q_values(
model_out_t, states_in_t["twin_q"], seq_lens, policy_t)[0]
q_t_det_policy = torch.min(q_t_det_policy, twin_q_t_det_policy)
# Target q network evaluation.
q_tp1 = target_model.get_q_values(target_model_out_tp1,
target_states_in_tp1["q"], seq_lens,
policy_tp1)[0]
if policy.config["twin_q"]:
twin_q_tp1 = target_model.get_twin_q_values(
target_model_out_tp1, target_states_in_tp1["twin_q"], seq_lens,
policy_tp1)[0]
# Take min over both twin-NNs.
q_tp1 = torch.min(q_tp1, twin_q_tp1)
q_t_selected = torch.squeeze(q_t, dim=-1)
if policy.config["twin_q"]:
twin_q_t_selected = torch.squeeze(twin_q_t, dim=-1)
q_tp1 -= alpha * log_pis_tp1
q_tp1_best = torch.squeeze(input=q_tp1, dim=-1)
q_tp1_best_masked = \
(1.0 - train_batch[SampleBatch.DONES].float()) * q_tp1_best
# compute RHS of bellman equation
q_t_selected_target = (train_batch[SampleBatch.REWARDS] +
(policy.config["gamma"]**policy.config["n_step"]) *
q_tp1_best_masked).detach()
# BURNIN #
B = state_batches[0].shape[0]
T = q_t_selected.shape[0] // B
seq_mask = sequence_mask(train_batch[SampleBatch.SEQ_LENS], T)
# Mask away also the burn-in sequence at the beginning.
burn_in = policy.config["burn_in"]
if burn_in > 0 and burn_in < T:
seq_mask[:, :burn_in] = False
seq_mask = seq_mask.reshape(-1)
num_valid = torch.sum(seq_mask)
def reduce_mean_valid(t):
return torch.sum(t[seq_mask]) / num_valid
# Compute the TD-error (potentially clipped).
base_td_error = torch.abs(q_t_selected - q_t_selected_target)
if policy.config["twin_q"]:
twin_td_error = torch.abs(twin_q_t_selected - q_t_selected_target)
td_error = 0.5 * (base_td_error + twin_td_error)
else:
td_error = base_td_error
critic_loss = [
reduce_mean_valid(train_batch[PRIO_WEIGHTS] *
huber_loss(base_td_error))
]
if policy.config["twin_q"]:
critic_loss.append(
reduce_mean_valid(train_batch[PRIO_WEIGHTS] *
huber_loss(twin_td_error)))
td_error = td_error * seq_mask
# Alpha- and actor losses.
# Note: In the papers, alpha is used directly, here we take the log.
# Discrete case: Multiply the action probs as weights with the original
# loss terms (no expectations needed).
if model.discrete:
weighted_log_alpha_loss = policy_t.detach() * (
-model.log_alpha * (log_pis_t + model.target_entropy).detach())
# Sum up weighted terms and mean over all batch items.
alpha_loss = reduce_mean_valid(
torch.sum(weighted_log_alpha_loss, dim=-1))
# Actor loss.
actor_loss = reduce_mean_valid(
torch.sum(
torch.mul(
# NOTE: No stop_grad around policy output here
# (compare with q_t_det_policy for continuous case).
policy_t,
alpha.detach() * log_pis_t - q_t.detach()),
dim=-1))
else:
alpha_loss = -reduce_mean_valid(
model.log_alpha * (log_pis_t + model.target_entropy).detach())
# Note: Do not detach q_t_det_policy here b/c is depends partly
# on the policy vars (policy sample pushed through Q-net).
# However, we must make sure `actor_loss` is not used to update
# the Q-net(s)' variables.
actor_loss = reduce_mean_valid(alpha.detach() * log_pis_t -
q_t_det_policy)
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["q_t"] = q_t * seq_mask[..., None]
model.tower_stats["policy_t"] = policy_t * seq_mask[..., None]
model.tower_stats["log_pis_t"] = log_pis_t * seq_mask[..., None]
model.tower_stats["actor_loss"] = actor_loss
model.tower_stats["critic_loss"] = critic_loss
model.tower_stats["alpha_loss"] = alpha_loss
# Store per time chunk (b/c we need only one mean
# prioritized replay weight per stored sequence).
model.tower_stats["td_error"] = torch.mean(td_error.reshape([-1, T]),
dim=-1)
# Return all loss terms corresponding to our optimizers.
return tuple([actor_loss] + critic_loss + [alpha_loss])
def ppo_surrogate_loss(
policy: Policy, model: ModelV2, dist_class: Type[TFActionDistribution],
train_batch: SampleBatch) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss for Proximal Policy Objective.
Args:
policy (Policy): The Policy to calculate the loss for.
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[ActionDistribution]: The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
logits, state = model.from_batch(train_batch)
curr_action_dist = dist_class(logits, model)
# RNN case: Mask away 0-padded chunks at end of time axis.
if state:
# Derive max_seq_len from the data itself, not from the seq_lens
# tensor. This is in case e.g. seq_lens=[2, 3], but the data is still
# 0-padded up to T=5 (as it's the case for attention nets).
B = tf.shape(train_batch["seq_lens"])[0]
max_seq_len = tf.shape(logits)[0] // B
mask = tf.sequence_mask(train_batch["seq_lens"], max_seq_len)
mask = tf.reshape(mask, [-1])
def reduce_mean_valid(t):
return tf.reduce_mean(tf.boolean_mask(t, mask))
# non-RNN case: No masking.
else:
mask = None
reduce_mean_valid = tf.reduce_mean
prev_action_dist = dist_class(train_batch[SampleBatch.ACTION_DIST_INPUTS],
model)
logp_ratio = tf.exp(
curr_action_dist.logp(train_batch[SampleBatch.ACTIONS]) -
train_batch[SampleBatch.ACTION_LOGP])
action_kl = prev_action_dist.kl(curr_action_dist)
mean_kl = reduce_mean_valid(action_kl)
curr_entropy = curr_action_dist.entropy()
mean_entropy = reduce_mean_valid(curr_entropy)
surrogate_loss = tf.minimum(
train_batch[Postprocessing.ADVANTAGES] * logp_ratio,
train_batch[Postprocessing.ADVANTAGES] * tf.clip_by_value(
logp_ratio, 1 - policy.config["clip_param"],
1 + policy.config["clip_param"]))
mean_policy_loss = reduce_mean_valid(-surrogate_loss)
if policy.config["use_gae"]:
prev_value_fn_out = train_batch[SampleBatch.VF_PREDS]
value_fn_out = model.value_function()
vf_loss1 = tf.math.square(value_fn_out -
train_batch[Postprocessing.VALUE_TARGETS])
vf_clipped = prev_value_fn_out + tf.clip_by_value(
value_fn_out - prev_value_fn_out, -policy.config["vf_clip_param"],
policy.config["vf_clip_param"])
vf_loss2 = tf.math.square(vf_clipped -
train_batch[Postprocessing.VALUE_TARGETS])
vf_loss = tf.maximum(vf_loss1, vf_loss2)
mean_vf_loss = reduce_mean_valid(vf_loss)
total_loss = reduce_mean_valid(
-surrogate_loss + policy.kl_coeff * action_kl +
policy.config["vf_loss_coeff"] * vf_loss -
policy.entropy_coeff * curr_entropy)
else:
mean_vf_loss = tf.constant(0.0)
total_loss = reduce_mean_valid(-surrogate_loss +
policy.kl_coeff * action_kl -
policy.entropy_coeff * curr_entropy)
# Store stats in policy for stats_fn.
policy._total_loss = total_loss
policy._mean_policy_loss = mean_policy_loss
policy._mean_vf_loss = mean_vf_loss
policy._mean_entropy = mean_entropy
policy._mean_kl = mean_kl
return total_loss
def model_value_predictions(
policy: Policy, input_dict: Dict[str, TensorType], state_batches,
model: ModelV2,
action_dist: ActionDistribution) -> Dict[str, TensorType]:
return {SampleBatch.VF_PREDS: model.value_function()}
def loss(
self,
model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch,
) -> List[TensorType]:
target_model = self.target_models[model]
twin_q = self.config["twin_q"]
gamma = self.config["gamma"]
n_step = self.config["n_step"]
use_huber = self.config["use_huber"]
huber_threshold = self.config["huber_threshold"]
l2_reg = self.config["l2_reg"]
input_dict = SampleBatch(
obs=train_batch[SampleBatch.CUR_OBS], _is_training=True
)
input_dict_next = SampleBatch(
obs=train_batch[SampleBatch.NEXT_OBS], _is_training=True
)
model_out_t, _ = model(input_dict, [], None)
model_out_tp1, _ = model(input_dict_next, [], None)
target_model_out_tp1, _ = target_model(input_dict_next, [], None)
# Policy network evaluation.
# prev_update_ops = set(tf1.get_collection(tf.GraphKeys.UPDATE_OPS))
policy_t = model.get_policy_output(model_out_t)
# policy_batchnorm_update_ops = list(
# set(tf1.get_collection(tf.GraphKeys.UPDATE_OPS)) - prev_update_ops)
policy_tp1 = target_model.get_policy_output(target_model_out_tp1)
# Action outputs.
if self.config["smooth_target_policy"]:
target_noise_clip = self.config["target_noise_clip"]
clipped_normal_sample = torch.clamp(
torch.normal(
mean=torch.zeros(policy_tp1.size()), std=self.config["target_noise"]
).to(policy_tp1.device),
-target_noise_clip,
target_noise_clip,
)
policy_tp1_smoothed = torch.min(
torch.max(
policy_tp1 + clipped_normal_sample,
torch.tensor(
self.action_space.low,
dtype=torch.float32,
device=policy_tp1.device,
),
),
torch.tensor(
self.action_space.high,
dtype=torch.float32,
device=policy_tp1.device,
),
)
else:
# No smoothing, just use deterministic actions.
policy_tp1_smoothed = policy_tp1
# Q-net(s) evaluation.
# prev_update_ops = set(tf1.get_collection(tf.GraphKeys.UPDATE_OPS))
# Q-values for given actions & observations in given current
q_t = model.get_q_values(model_out_t, train_batch[SampleBatch.ACTIONS])
# Q-values for current policy (no noise) in given current state
q_t_det_policy = model.get_q_values(model_out_t, policy_t)
actor_loss = -torch.mean(q_t_det_policy)
if twin_q:
twin_q_t = model.get_twin_q_values(
model_out_t, train_batch[SampleBatch.ACTIONS]
)
# q_batchnorm_update_ops = list(
# set(tf1.get_collection(tf.GraphKeys.UPDATE_OPS)) - prev_update_ops)
# Target q-net(s) evaluation.
q_tp1 = target_model.get_q_values(target_model_out_tp1, policy_tp1_smoothed)
if twin_q:
twin_q_tp1 = target_model.get_twin_q_values(
target_model_out_tp1, policy_tp1_smoothed
)
q_t_selected = torch.squeeze(q_t, axis=len(q_t.shape) - 1)
if twin_q:
twin_q_t_selected = torch.squeeze(twin_q_t, axis=len(q_t.shape) - 1)
q_tp1 = torch.min(q_tp1, twin_q_tp1)
q_tp1_best = torch.squeeze(input=q_tp1, axis=len(q_tp1.shape) - 1)
q_tp1_best_masked = (1.0 - train_batch[SampleBatch.DONES].float()) * q_tp1_best
# Compute RHS of bellman equation.
q_t_selected_target = (
train_batch[SampleBatch.REWARDS] + gamma ** n_step * q_tp1_best_masked
).detach()
# Compute the error (potentially clipped).
if twin_q:
td_error = q_t_selected - q_t_selected_target
twin_td_error = twin_q_t_selected - q_t_selected_target
if use_huber:
errors = huber_loss(td_error, huber_threshold) + huber_loss(
twin_td_error, huber_threshold
)
else:
errors = 0.5 * (
torch.pow(td_error, 2.0) + torch.pow(twin_td_error, 2.0)
)
else:
td_error = q_t_selected - q_t_selected_target
if use_huber:
errors = huber_loss(td_error, huber_threshold)
else:
errors = 0.5 * torch.pow(td_error, 2.0)
critic_loss = torch.mean(train_batch[PRIO_WEIGHTS] * errors)
# Add l2-regularization if required.
if l2_reg is not None:
for name, var in model.policy_variables(as_dict=True).items():
if "bias" not in name:
actor_loss += l2_reg * l2_loss(var)
for name, var in model.q_variables(as_dict=True).items():
if "bias" not in name:
critic_loss += l2_reg * l2_loss(var)
# Model self-supervised losses.
if self.config["use_state_preprocessor"]:
# Expand input_dict in case custom_loss' need them.
input_dict[SampleBatch.ACTIONS] = train_batch[SampleBatch.ACTIONS]
input_dict[SampleBatch.REWARDS] = train_batch[SampleBatch.REWARDS]
input_dict[SampleBatch.DONES] = train_batch[SampleBatch.DONES]
input_dict[SampleBatch.NEXT_OBS] = train_batch[SampleBatch.NEXT_OBS]
[actor_loss, critic_loss] = model.custom_loss(
[actor_loss, critic_loss], input_dict
)
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["q_t"] = q_t
model.tower_stats["actor_loss"] = actor_loss
model.tower_stats["critic_loss"] = critic_loss
# TD-error tensor in final stats
# will be concatenated and retrieved for each individual batch item.
model.tower_stats["td_error"] = td_error
# Return two loss terms (corresponding to the two optimizers, we create).
return [actor_loss, critic_loss]
def _get_q_value(self, model: ModelV2, model_out: TensorType,
actions: TensorType) -> TensorType:
# helper function to compute the pessimistic q value
q1 = model.get_q_values(model_out, actions)
q2 = model.get_twin_q_values(model_out, actions)
return torch.minimum(q1, q2)
def appo_surrogate_loss(
policy: Policy, model: ModelV2, dist_class: Type[TFActionDistribution],
train_batch: SampleBatch) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss for APPO.
With IS modifications and V-trace for Advantage Estimation.
Args:
policy (Policy): The Policy to calculate the loss for.
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[ActionDistribution]): The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
model_out, _ = model.from_batch(train_batch)
action_dist = dist_class(model_out, model)
if isinstance(policy.action_space, gym.spaces.Discrete):
is_multidiscrete = False
output_hidden_shape = [policy.action_space.n]
elif isinstance(policy.action_space,
gym.spaces.multi_discrete.MultiDiscrete):
is_multidiscrete = True
output_hidden_shape = policy.action_space.nvec.astype(np.int32)
else:
is_multidiscrete = False
output_hidden_shape = 1
# TODO: (sven) deprecate this when trajectory view API gets activated.
def make_time_major(*args, **kw):
return _make_time_major(policy, train_batch.get("seq_lens"), *args,
**kw)
actions = train_batch[SampleBatch.ACTIONS]
dones = train_batch[SampleBatch.DONES]
rewards = train_batch[SampleBatch.REWARDS]
behaviour_logits = train_batch[SampleBatch.ACTION_DIST_INPUTS]
target_model_out, _ = policy.target_model.from_batch(train_batch)
prev_action_dist = dist_class(behaviour_logits, policy.model)
values = policy.model.value_function()
values_time_major = make_time_major(values)
policy.model_vars = policy.model.variables()
policy.target_model_vars = policy.target_model.variables()
if policy.is_recurrent():
max_seq_len = tf.reduce_max(train_batch["seq_lens"]) - 1
mask = tf.sequence_mask(train_batch["seq_lens"], max_seq_len)
mask = tf.reshape(mask, [-1])
mask = make_time_major(mask, drop_last=policy.config["vtrace"])
def reduce_mean_valid(t):
return tf.reduce_mean(tf.boolean_mask(t, mask))
else:
reduce_mean_valid = tf.reduce_mean
if policy.config["vtrace"]:
logger.debug("Using V-Trace surrogate loss (vtrace=True)")
# Prepare actions for loss.
loss_actions = actions if is_multidiscrete else tf.expand_dims(actions,
axis=1)
old_policy_behaviour_logits = tf.stop_gradient(target_model_out)
old_policy_action_dist = dist_class(old_policy_behaviour_logits, model)
# Prepare KL for Loss
mean_kl = make_time_major(old_policy_action_dist.multi_kl(action_dist),
drop_last=True)
unpacked_behaviour_logits = tf.split(behaviour_logits,
output_hidden_shape,
axis=1)
unpacked_old_policy_behaviour_logits = tf.split(
old_policy_behaviour_logits, output_hidden_shape, axis=1)
# Compute vtrace on the CPU for better perf.
with tf.device("/cpu:0"):
vtrace_returns = vtrace.multi_from_logits(
behaviour_policy_logits=make_time_major(
unpacked_behaviour_logits, drop_last=True),
target_policy_logits=make_time_major(
unpacked_old_policy_behaviour_logits, drop_last=True),
actions=tf.unstack(make_time_major(loss_actions,
drop_last=True),
axis=2),
discounts=tf.cast(
~make_time_major(tf.cast(dones, tf.bool), drop_last=True),
tf.float32) * policy.config["gamma"],
rewards=make_time_major(rewards, drop_last=True),
values=values_time_major[:-1], # drop-last=True
bootstrap_value=values_time_major[-1],
dist_class=Categorical if is_multidiscrete else dist_class,
model=model,
clip_rho_threshold=tf.cast(
policy.config["vtrace_clip_rho_threshold"], tf.float32),
clip_pg_rho_threshold=tf.cast(
policy.config["vtrace_clip_pg_rho_threshold"], tf.float32),
)
actions_logp = make_time_major(action_dist.logp(actions),
drop_last=True)
prev_actions_logp = make_time_major(prev_action_dist.logp(actions),
drop_last=True)
old_policy_actions_logp = make_time_major(
old_policy_action_dist.logp(actions), drop_last=True)
is_ratio = tf.clip_by_value(
tf.math.exp(prev_actions_logp - old_policy_actions_logp), 0.0, 2.0)
logp_ratio = is_ratio * tf.exp(actions_logp - prev_actions_logp)
policy._is_ratio = is_ratio
advantages = vtrace_returns.pg_advantages
surrogate_loss = tf.minimum(
advantages * logp_ratio,
advantages *
tf.clip_by_value(logp_ratio, 1 - policy.config["clip_param"],
1 + policy.config["clip_param"]))
action_kl = tf.reduce_mean(mean_kl, axis=0) \
if is_multidiscrete else mean_kl
mean_kl = reduce_mean_valid(action_kl)
mean_policy_loss = -reduce_mean_valid(surrogate_loss)
# The value function loss.
delta = values_time_major[:-1] - vtrace_returns.vs
value_targets = vtrace_returns.vs
mean_vf_loss = 0.5 * reduce_mean_valid(tf.math.square(delta))
# The entropy loss.
actions_entropy = make_time_major(action_dist.multi_entropy(),
drop_last=True)
mean_entropy = reduce_mean_valid(actions_entropy)
else:
logger.debug("Using PPO surrogate loss (vtrace=False)")
# Prepare KL for Loss
mean_kl = make_time_major(prev_action_dist.multi_kl(action_dist))
logp_ratio = tf.math.exp(
make_time_major(action_dist.logp(actions)) -
make_time_major(prev_action_dist.logp(actions)))
advantages = make_time_major(train_batch[Postprocessing.ADVANTAGES])
surrogate_loss = tf.minimum(
advantages * logp_ratio,
advantages *
tf.clip_by_value(logp_ratio, 1 - policy.config["clip_param"],
1 + policy.config["clip_param"]))
action_kl = tf.reduce_mean(mean_kl, axis=0) \
if is_multidiscrete else mean_kl
mean_kl = reduce_mean_valid(action_kl)
mean_policy_loss = -reduce_mean_valid(surrogate_loss)
# The value function loss.
value_targets = make_time_major(
train_batch[Postprocessing.VALUE_TARGETS])
delta = values_time_major - value_targets
mean_vf_loss = 0.5 * reduce_mean_valid(tf.math.square(delta))
# The entropy loss.
mean_entropy = reduce_mean_valid(
make_time_major(action_dist.multi_entropy()))
# The summed weighted loss
total_loss = mean_policy_loss + \
mean_vf_loss * policy.config["vf_loss_coeff"] - \
mean_entropy * policy.config["entropy_coeff"]
# Optional additional KL Loss
if policy.config["use_kl_loss"]:
total_loss += policy.kl_coeff * mean_kl
policy._total_loss = total_loss
policy._mean_policy_loss = mean_policy_loss
policy._mean_kl = mean_kl
policy._mean_vf_loss = mean_vf_loss
policy._mean_entropy = mean_entropy
policy._value_targets = value_targets
# Store stats in policy for stats_fn.
return total_loss
def _compute_adv_and_logps(
self,
model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch,
) -> None:
# uses mean|max to compute estimate of advantages
# continuous/discrete action spaces:
# A(s_t, a_t) = Q(s_t, a_t) - max_{a^j} Q(s_t, a^j)
# where a^j is m times sampled from the policy p(a | s_t)
# A(s_t, a_t) = Q(s_t, a_t) - avg( Q(s_t, a^j) )
# where a^j is m times sampled from the policy p(a | s_t)
# questions: Do we use pessimistic q approximate or the normal one?
advantage_type = self.config["advantage_type"]
n_action_sample = self.config["n_action_sample"]
batch_size = len(train_batch)
out_t, _ = model(train_batch)
# construct pi(s_t) for sampling actions
pi_s_t = dist_class(model.get_policy_output(out_t), model)
policy_actions = pi_s_t.dist.sample((n_action_sample, )) # samples
if self._is_action_discrete:
flat_actions = policy_actions.reshape(-1)
else:
flat_actions = policy_actions.reshape(-1, *self.action_space.shape)
# compute the logp of the actions in the dataset (for computing actor's loss)
action_logp = pi_s_t.dist.log_prob(train_batch[SampleBatch.ACTIONS])
# fix the shape if it's not canonical (i.e. shape[-1] != 1)
if len(action_logp.shape) <= 1:
action_logp.unsqueeze_(-1)
train_batch[SampleBatch.ACTION_LOGP] = action_logp
reshaped_s_t = train_batch[SampleBatch.OBS].view(
1, batch_size, *self.observation_space.shape)
reshaped_s_t = reshaped_s_t.expand(n_action_sample, batch_size,
*self.observation_space.shape)
flat_s_t = reshaped_s_t.reshape(-1, *self.observation_space.shape)
input_v_t = SampleBatch(**{
SampleBatch.OBS: flat_s_t,
SampleBatch.ACTIONS: flat_actions
})
out_v_t, _ = model(input_v_t)
flat_q_st_pi = self._get_q_value(model, out_v_t, flat_actions)
reshaped_q_st_pi = flat_q_st_pi.reshape(-1, batch_size, 1)
if advantage_type == "mean":
v_t = reshaped_q_st_pi.mean(dim=0)
elif advantage_type == "max":
v_t, _ = reshaped_q_st_pi.max(dim=0)
else:
raise ValueError(f"Invalid advantage type: {advantage_type}.")
q_t = self._get_q_value(model, out_t, train_batch[SampleBatch.ACTIONS])
adv_t = q_t - v_t
train_batch["advantages"] = adv_t
# logging
self.log("q_batch_avg", q_t.mean())
self.log("q_batch_max", q_t.max())
self.log("q_batch_min", q_t.min())
self.log("v_batch_avg", v_t.mean())
self.log("v_batch_max", v_t.max())
self.log("v_batch_min", v_t.min())
self.log("adv_batch_avg", adv_t.mean())
self.log("adv_batch_max", adv_t.max())
self.log("adv_batch_min", adv_t.min())
self.log("reward_batch_avg", train_batch[SampleBatch.REWARDS].mean())
def sac_actor_critic_loss(
policy: Policy, model: ModelV2, dist_class: Type[TFActionDistribution],
train_batch: SampleBatch) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss for the Soft Actor Critic.
Args:
policy (Policy): The Policy to calculate the loss for.
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[ActionDistribution]: The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
# Should be True only for debugging purposes (e.g. test cases)!
deterministic = policy.config["_deterministic_loss"]
# Get the base model output from the train batch.
model_out_t, _ = model({
"obs": train_batch[SampleBatch.CUR_OBS],
"is_training": policy._get_is_training_placeholder(),
}, [], None)
# Get the base model output from the next observations in the train batch.
model_out_tp1, _ = model({
"obs": train_batch[SampleBatch.NEXT_OBS],
"is_training": policy._get_is_training_placeholder(),
}, [], None)
# Get the target model's base outputs from the next observations in the
# train batch.
target_model_out_tp1, _ = policy.target_model({
"obs": train_batch[SampleBatch.NEXT_OBS],
"is_training": policy._get_is_training_placeholder(),
}, [], None)
# Discrete actions case.
if model.discrete:
# Get all action probs directly from pi and form their logp.
log_pis_t = tf.nn.log_softmax(model.get_policy_output(model_out_t), -1)
policy_t = tf.math.exp(log_pis_t)
log_pis_tp1 = tf.nn.log_softmax(
model.get_policy_output(model_out_tp1), -1)
policy_tp1 = tf.math.exp(log_pis_tp1)
# Q-values.
q_t = model.get_q_values(model_out_t)
# Target Q-values.
q_tp1 = policy.target_model.get_q_values(target_model_out_tp1)
if policy.config["twin_q"]:
twin_q_t = model.get_twin_q_values(model_out_t)
twin_q_tp1 = policy.target_model.get_twin_q_values(
target_model_out_tp1)
q_tp1 = tf.reduce_min((q_tp1, twin_q_tp1), axis=0)
q_tp1 -= model.alpha * log_pis_tp1
# Actually selected Q-values (from the actions batch).
one_hot = tf.one_hot(
train_batch[SampleBatch.ACTIONS], depth=q_t.shape.as_list()[-1])
q_t_selected = tf.reduce_sum(q_t * one_hot, axis=-1)
if policy.config["twin_q"]:
twin_q_t_selected = tf.reduce_sum(twin_q_t * one_hot, axis=-1)
# Discrete case: "Best" means weighted by the policy (prob) outputs.
q_tp1_best = tf.reduce_sum(tf.multiply(policy_tp1, q_tp1), axis=-1)
q_tp1_best_masked = \
(1.0 - tf.cast(train_batch[SampleBatch.DONES], tf.float32)) * \
q_tp1_best
# Continuous actions case.
else:
# Sample simgle actions from distribution.
action_dist_class = _get_dist_class(policy.config, policy.action_space)
action_dist_t = action_dist_class(
model.get_policy_output(model_out_t), policy.model)
policy_t = action_dist_t.sample() if not deterministic else \
action_dist_t.deterministic_sample()
log_pis_t = tf.expand_dims(action_dist_t.logp(policy_t), -1)
action_dist_tp1 = action_dist_class(
model.get_policy_output(model_out_tp1), policy.model)
policy_tp1 = action_dist_tp1.sample() if not deterministic else \
action_dist_tp1.deterministic_sample()
log_pis_tp1 = tf.expand_dims(action_dist_tp1.logp(policy_tp1), -1)
# Q-values for the actually selected actions.
q_t = model.get_q_values(model_out_t, train_batch[SampleBatch.ACTIONS])
if policy.config["twin_q"]:
twin_q_t = model.get_twin_q_values(
model_out_t, train_batch[SampleBatch.ACTIONS])
# Q-values for current policy in given current state.
q_t_det_policy = model.get_q_values(model_out_t, policy_t)
if policy.config["twin_q"]:
twin_q_t_det_policy = model.get_twin_q_values(
model_out_t, policy_t)
q_t_det_policy = tf.reduce_min(
(q_t_det_policy, twin_q_t_det_policy), axis=0)
# target q network evaluation
q_tp1 = policy.target_model.get_q_values(target_model_out_tp1,
policy_tp1)
if policy.config["twin_q"]:
twin_q_tp1 = policy.target_model.get_twin_q_values(
target_model_out_tp1, policy_tp1)
# Take min over both twin-NNs.
q_tp1 = tf.reduce_min((q_tp1, twin_q_tp1), axis=0)
q_t_selected = tf.squeeze(q_t, axis=len(q_t.shape) - 1)
if policy.config["twin_q"]:
twin_q_t_selected = tf.squeeze(twin_q_t, axis=len(q_t.shape) - 1)
q_tp1 -= model.alpha * log_pis_tp1
q_tp1_best = tf.squeeze(input=q_tp1, axis=len(q_tp1.shape) - 1)
q_tp1_best_masked = (1.0 - tf.cast(train_batch[SampleBatch.DONES],
tf.float32)) * q_tp1_best
# Compute RHS of bellman equation for the Q-loss (critic(s)).
q_t_selected_target = tf.stop_gradient(
train_batch[SampleBatch.REWARDS] +
policy.config["gamma"]**policy.config["n_step"] * q_tp1_best_masked)
# Compute the TD-error (potentially clipped).
base_td_error = tf.math.abs(q_t_selected - q_t_selected_target)
if policy.config["twin_q"]:
twin_td_error = tf.math.abs(twin_q_t_selected - q_t_selected_target)
td_error = 0.5 * (base_td_error + twin_td_error)
else:
td_error = base_td_error
# Calculate one or two critic losses (2 in the twin_q case).
prio_weights = tf.cast(train_batch[PRIO_WEIGHTS], tf.float32)
critic_loss = [tf.reduce_mean(prio_weights * huber_loss(base_td_error))]
if policy.config["twin_q"]:
critic_loss.append(
tf.reduce_mean(prio_weights * huber_loss(twin_td_error)))
# Alpha- and actor losses.
# Note: In the papers, alpha is used directly, here we take the log.
# Discrete case: Multiply the action probs as weights with the original
# loss terms (no expectations needed).
if model.discrete:
alpha_loss = tf.reduce_mean(
tf.reduce_sum(
tf.multiply(
tf.stop_gradient(policy_t), -model.log_alpha *
tf.stop_gradient(log_pis_t + model.target_entropy)),
axis=-1))
actor_loss = tf.reduce_mean(
tf.reduce_sum(
tf.multiply(
# NOTE: No stop_grad around policy output here
# (compare with q_t_det_policy for continuous case).
policy_t,
model.alpha * log_pis_t - tf.stop_gradient(q_t)),
axis=-1))
else:
alpha_loss = -tf.reduce_mean(
model.log_alpha *
tf.stop_gradient(log_pis_t + model.target_entropy))
actor_loss = tf.reduce_mean(model.alpha * log_pis_t - q_t_det_policy)
# Save for stats function.
policy.policy_t = policy_t
policy.q_t = q_t
policy.td_error = td_error
policy.actor_loss = actor_loss
policy.critic_loss = critic_loss
policy.alpha_loss = alpha_loss
policy.alpha_value = model.alpha
policy.target_entropy = model.target_entropy
# In a custom apply op we handle the losses separately, but return them
# combined in one loss here.
return actor_loss + tf.math.add_n(critic_loss) + alpha_loss
def ddpg_actor_critic_loss(policy: Policy, model: ModelV2, _,
train_batch: SampleBatch) -> TensorType:
twin_q = policy.config["twin_q"]
gamma = policy.config["gamma"]
n_step = policy.config["n_step"]
use_huber = policy.config["use_huber"]
huber_threshold = policy.config["huber_threshold"]
l2_reg = policy.config["l2_reg"]
input_dict = SampleBatch(obs=train_batch[SampleBatch.CUR_OBS],
_is_training=True)
input_dict_next = SampleBatch(obs=train_batch[SampleBatch.NEXT_OBS],
_is_training=True)
model_out_t, _ = model(input_dict, [], None)
model_out_tp1, _ = model(input_dict_next, [], None)
target_model_out_tp1, _ = policy.target_model(input_dict_next, [], None)
policy.target_q_func_vars = policy.target_model.variables()
# Policy network evaluation.
policy_t = model.get_policy_output(model_out_t)
policy_tp1 = policy.target_model.get_policy_output(target_model_out_tp1)
# Action outputs.
if policy.config["smooth_target_policy"]:
target_noise_clip = policy.config["target_noise_clip"]
clipped_normal_sample = tf.clip_by_value(
tf.random.normal(tf.shape(policy_tp1),
stddev=policy.config["target_noise"]),
-target_noise_clip,
target_noise_clip,
)
policy_tp1_smoothed = tf.clip_by_value(
policy_tp1 + clipped_normal_sample,
policy.action_space.low * tf.ones_like(policy_tp1),
policy.action_space.high * tf.ones_like(policy_tp1),
)
else:
# No smoothing, just use deterministic actions.
policy_tp1_smoothed = policy_tp1
# Q-net(s) evaluation.
# prev_update_ops = set(tf.get_collection(tf.GraphKeys.UPDATE_OPS))
# Q-values for given actions & observations in given current
q_t = model.get_q_values(model_out_t, train_batch[SampleBatch.ACTIONS])
# Q-values for current policy (no noise) in given current state
q_t_det_policy = model.get_q_values(model_out_t, policy_t)
if twin_q:
twin_q_t = model.get_twin_q_values(model_out_t,
train_batch[SampleBatch.ACTIONS])
# Target q-net(s) evaluation.
q_tp1 = policy.target_model.get_q_values(target_model_out_tp1,
policy_tp1_smoothed)
if twin_q:
twin_q_tp1 = policy.target_model.get_twin_q_values(
target_model_out_tp1, policy_tp1_smoothed)
q_t_selected = tf.squeeze(q_t, axis=len(q_t.shape) - 1)
if twin_q:
twin_q_t_selected = tf.squeeze(twin_q_t, axis=len(q_t.shape) - 1)
q_tp1 = tf.minimum(q_tp1, twin_q_tp1)
q_tp1_best = tf.squeeze(input=q_tp1, axis=len(q_tp1.shape) - 1)
q_tp1_best_masked = (
1.0 - tf.cast(train_batch[SampleBatch.DONES], tf.float32)) * q_tp1_best
# Compute RHS of bellman equation.
q_t_selected_target = tf.stop_gradient(
tf.cast(train_batch[SampleBatch.REWARDS], tf.float32) +
gamma**n_step * q_tp1_best_masked)
# Compute the error (potentially clipped).
if twin_q:
td_error = q_t_selected - q_t_selected_target
twin_td_error = twin_q_t_selected - q_t_selected_target
if use_huber:
errors = huber_loss(td_error, huber_threshold) + huber_loss(
twin_td_error, huber_threshold)
else:
errors = 0.5 * tf.math.square(td_error) + 0.5 * tf.math.square(
twin_td_error)
else:
td_error = q_t_selected - q_t_selected_target
if use_huber:
errors = huber_loss(td_error, huber_threshold)
else:
errors = 0.5 * tf.math.square(td_error)
critic_loss = tf.reduce_mean(
tf.cast(train_batch[PRIO_WEIGHTS], tf.float32) * errors)
actor_loss = -tf.reduce_mean(q_t_det_policy)
# Add l2-regularization if required.
if l2_reg is not None:
for var in policy.model.policy_variables():
if "bias" not in var.name:
actor_loss += l2_reg * tf.nn.l2_loss(var)
for var in policy.model.q_variables():
if "bias" not in var.name:
critic_loss += l2_reg * tf.nn.l2_loss(var)
# Model self-supervised losses.
if policy.config["use_state_preprocessor"]:
# Expand input_dict in case custom_loss' need them.
input_dict[SampleBatch.ACTIONS] = train_batch[SampleBatch.ACTIONS]
input_dict[SampleBatch.REWARDS] = train_batch[SampleBatch.REWARDS]
input_dict[SampleBatch.DONES] = train_batch[SampleBatch.DONES]
input_dict[SampleBatch.NEXT_OBS] = train_batch[SampleBatch.NEXT_OBS]
if log_once("ddpg_custom_loss"):
logger.warning(
"You are using a state-preprocessor with DDPG and "
"therefore, `custom_loss` will be called on your Model! "
"Please be aware that DDPG now uses the ModelV2 API, which "
"merges all previously separate sub-models (policy_model, "
"q_model, and twin_q_model) into one ModelV2, on which "
"`custom_loss` is called, passing it "
"[actor_loss, critic_loss] as 1st argument. "
"You may have to change your custom loss function to handle "
"this.")
[actor_loss,
critic_loss] = model.custom_loss([actor_loss, critic_loss],
input_dict)
# Store values for stats function.
policy.actor_loss = actor_loss
policy.critic_loss = critic_loss
policy.td_error = td_error
policy.q_t = q_t
# Return one loss value (even though we treat them separately in our
# 2 optimizers: actor and critic).
return policy.critic_loss + policy.actor_loss
def context(self) -> contextlib.AbstractContextManager:
"""Returns a contextmanager for the current TF graph."""
if self.graph:
return self.graph.as_default()
else:
return ModelV2.context(self)
def loss(
self,
model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch,
) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss function.
Args:
model: The Model to calculate the loss for.
dist_class: The action distr. class.
train_batch: The training data.
Returns:
The PPO loss tensor given the input batch.
"""
logits, state = model(train_batch)
self.cur_lr = self.config["lr"]
if self.config["worker_index"]:
self.loss_obj = WorkerLoss(
model=model,
dist_class=dist_class,
actions=train_batch[SampleBatch.ACTIONS],
curr_logits=logits,
behaviour_logits=train_batch[SampleBatch.ACTION_DIST_INPUTS],
advantages=train_batch[Postprocessing.ADVANTAGES],
value_fn=model.value_function(),
value_targets=train_batch[Postprocessing.VALUE_TARGETS],
vf_preds=train_batch[SampleBatch.VF_PREDS],
cur_kl_coeff=0.0,
entropy_coeff=self.config["entropy_coeff"],
clip_param=self.config["clip_param"],
vf_clip_param=self.config["vf_clip_param"],
vf_loss_coeff=self.config["vf_loss_coeff"],
clip_loss=False,
)
else:
self.var_list = model.named_parameters()
# `split` may not exist yet (during test-loss call), use a dummy value.
# Cannot use get here due to train_batch being a TrackingDict.
if "split" in train_batch:
split = train_batch["split"]
else:
split_shape = (
self.config["inner_adaptation_steps"],
self.config["num_workers"],
)
split_const = int(train_batch["obs"].shape[0] //
(split_shape[0] * split_shape[1]))
split = torch.ones(split_shape, dtype=int) * split_const
self.loss_obj = MAMLLoss(
model=model,
dist_class=dist_class,
value_targets=train_batch[Postprocessing.VALUE_TARGETS],
advantages=train_batch[Postprocessing.ADVANTAGES],
actions=train_batch[SampleBatch.ACTIONS],
behaviour_logits=train_batch[SampleBatch.ACTION_DIST_INPUTS],
vf_preds=train_batch[SampleBatch.VF_PREDS],
cur_kl_coeff=self.kl_coeff_val,
policy_vars=self.var_list,
obs=train_batch[SampleBatch.CUR_OBS],
num_tasks=self.config["num_workers"],
split=split,
config=self.config,
inner_adaptation_steps=self.config["inner_adaptation_steps"],
entropy_coeff=self.config["entropy_coeff"],
clip_param=self.config["clip_param"],
vf_clip_param=self.config["vf_clip_param"],
vf_loss_coeff=self.config["vf_loss_coeff"],
use_gae=self.config["use_gae"],
meta_opt=self.meta_opt,
)
return self.loss_obj.loss
def loss(
self,
model: ModelV2,
dist_class: Type[ActionDistribution],
train_batch: SampleBatch,
) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss for APPO.
With IS modifications and V-trace for Advantage Estimation.
Args:
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[ActionDistribution]): The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
target_model = self.target_models[model]
model_out, _ = model(train_batch)
action_dist = dist_class(model_out, model)
if isinstance(self.action_space, gym.spaces.Discrete):
is_multidiscrete = False
output_hidden_shape = [self.action_space.n]
elif isinstance(self.action_space, gym.spaces.multi_discrete.MultiDiscrete):
is_multidiscrete = True
output_hidden_shape = self.action_space.nvec.astype(np.int32)
else:
is_multidiscrete = False
output_hidden_shape = 1
def _make_time_major(*args, **kwargs):
return make_time_major(
self, train_batch.get(SampleBatch.SEQ_LENS), *args, **kwargs
)
actions = train_batch[SampleBatch.ACTIONS]
dones = train_batch[SampleBatch.DONES]
rewards = train_batch[SampleBatch.REWARDS]
behaviour_logits = train_batch[SampleBatch.ACTION_DIST_INPUTS]
target_model_out, _ = target_model(train_batch)
prev_action_dist = dist_class(behaviour_logits, model)
values = model.value_function()
values_time_major = _make_time_major(values)
drop_last = self.config["vtrace"] and self.config["vtrace_drop_last_ts"]
if self.is_recurrent():
max_seq_len = torch.max(train_batch[SampleBatch.SEQ_LENS])
mask = sequence_mask(train_batch[SampleBatch.SEQ_LENS], max_seq_len)
mask = torch.reshape(mask, [-1])
mask = _make_time_major(mask, drop_last=drop_last)
num_valid = torch.sum(mask)
def reduce_mean_valid(t):
return torch.sum(t[mask]) / num_valid
else:
reduce_mean_valid = torch.mean
if self.config["vtrace"]:
logger.debug(
"Using V-Trace surrogate loss (vtrace=True; " f"drop_last={drop_last})"
)
old_policy_behaviour_logits = target_model_out.detach()
old_policy_action_dist = dist_class(old_policy_behaviour_logits, model)
if isinstance(output_hidden_shape, (list, tuple, np.ndarray)):
unpacked_behaviour_logits = torch.split(
behaviour_logits, list(output_hidden_shape), dim=1
)
unpacked_old_policy_behaviour_logits = torch.split(
old_policy_behaviour_logits, list(output_hidden_shape), dim=1
)
else:
unpacked_behaviour_logits = torch.chunk(
behaviour_logits, output_hidden_shape, dim=1
)
unpacked_old_policy_behaviour_logits = torch.chunk(
old_policy_behaviour_logits, output_hidden_shape, dim=1
)
# Prepare actions for loss.
loss_actions = (
actions if is_multidiscrete else torch.unsqueeze(actions, dim=1)
)
# Prepare KL for loss.
action_kl = _make_time_major(
old_policy_action_dist.kl(action_dist), drop_last=drop_last
)
# Compute vtrace on the CPU for better perf.
vtrace_returns = vtrace.multi_from_logits(
behaviour_policy_logits=_make_time_major(
unpacked_behaviour_logits, drop_last=drop_last
),
target_policy_logits=_make_time_major(
unpacked_old_policy_behaviour_logits, drop_last=drop_last
),
actions=torch.unbind(
_make_time_major(loss_actions, drop_last=drop_last), dim=2
),
discounts=(1.0 - _make_time_major(dones, drop_last=drop_last).float())
* self.config["gamma"],
rewards=_make_time_major(rewards, drop_last=drop_last),
values=values_time_major[:-1] if drop_last else values_time_major,
bootstrap_value=values_time_major[-1],
dist_class=TorchCategorical if is_multidiscrete else dist_class,
model=model,
clip_rho_threshold=self.config["vtrace_clip_rho_threshold"],
clip_pg_rho_threshold=self.config["vtrace_clip_pg_rho_threshold"],
)
actions_logp = _make_time_major(
action_dist.logp(actions), drop_last=drop_last
)
prev_actions_logp = _make_time_major(
prev_action_dist.logp(actions), drop_last=drop_last
)
old_policy_actions_logp = _make_time_major(
old_policy_action_dist.logp(actions), drop_last=drop_last
)
is_ratio = torch.clamp(
torch.exp(prev_actions_logp - old_policy_actions_logp), 0.0, 2.0
)
logp_ratio = is_ratio * torch.exp(actions_logp - prev_actions_logp)
self._is_ratio = is_ratio
advantages = vtrace_returns.pg_advantages.to(logp_ratio.device)
surrogate_loss = torch.min(
advantages * logp_ratio,
advantages
* torch.clamp(
logp_ratio,
1 - self.config["clip_param"],
1 + self.config["clip_param"],
),
)
mean_kl_loss = reduce_mean_valid(action_kl)
mean_policy_loss = -reduce_mean_valid(surrogate_loss)
# The value function loss.
value_targets = vtrace_returns.vs.to(values_time_major.device)
if drop_last:
delta = values_time_major[:-1] - value_targets
else:
delta = values_time_major - value_targets
mean_vf_loss = 0.5 * reduce_mean_valid(torch.pow(delta, 2.0))
# The entropy loss.
mean_entropy = reduce_mean_valid(
_make_time_major(action_dist.entropy(), drop_last=drop_last)
)
else:
logger.debug("Using PPO surrogate loss (vtrace=False)")
# Prepare KL for Loss
action_kl = _make_time_major(prev_action_dist.kl(action_dist))
actions_logp = _make_time_major(action_dist.logp(actions))
prev_actions_logp = _make_time_major(prev_action_dist.logp(actions))
logp_ratio = torch.exp(actions_logp - prev_actions_logp)
advantages = _make_time_major(train_batch[Postprocessing.ADVANTAGES])
surrogate_loss = torch.min(
advantages * logp_ratio,
advantages
* torch.clamp(
logp_ratio,
1 - self.config["clip_param"],
1 + self.config["clip_param"],
),
)
mean_kl_loss = reduce_mean_valid(action_kl)
mean_policy_loss = -reduce_mean_valid(surrogate_loss)
# The value function loss.
value_targets = _make_time_major(train_batch[Postprocessing.VALUE_TARGETS])
delta = values_time_major - value_targets
mean_vf_loss = 0.5 * reduce_mean_valid(torch.pow(delta, 2.0))
# The entropy loss.
mean_entropy = reduce_mean_valid(_make_time_major(action_dist.entropy()))
# The summed weighted loss
total_loss = (
mean_policy_loss
+ mean_vf_loss * self.config["vf_loss_coeff"]
- mean_entropy * self.entropy_coeff
)
# Optional additional KL Loss
if self.config["use_kl_loss"]:
total_loss += self.kl_coeff * mean_kl_loss
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["total_loss"] = total_loss
model.tower_stats["mean_policy_loss"] = mean_policy_loss
model.tower_stats["mean_kl_loss"] = mean_kl_loss
model.tower_stats["mean_vf_loss"] = mean_vf_loss
model.tower_stats["mean_entropy"] = mean_entropy
model.tower_stats["value_targets"] = value_targets
model.tower_stats["vf_explained_var"] = explained_variance(
torch.reshape(value_targets, [-1]),
torch.reshape(
values_time_major[:-1] if drop_last else values_time_major, [-1]
),
)
return total_loss
def ppo_surrogate_loss(
policy: Policy, model: ModelV2,
dist_class: Type[TorchDistributionWrapper],
train_batch: SampleBatch) -> Union[TensorType, List[TensorType]]:
"""Constructs the loss for Proximal Policy Objective.
Args:
policy (Policy): The Policy to calculate the loss for.
model (ModelV2): The Model to calculate the loss for.
dist_class (Type[ActionDistribution]: The action distr. class.
train_batch (SampleBatch): The training data.
Returns:
Union[TensorType, List[TensorType]]: A single loss tensor or a list
of loss tensors.
"""
logits, state = model(train_batch)
curr_action_dist = dist_class(logits, model)
# RNN case: Mask away 0-padded chunks at end of time axis.
if state:
B = len(train_batch[SampleBatch.SEQ_LENS])
max_seq_len = logits.shape[0] // B
mask = sequence_mask(train_batch[SampleBatch.SEQ_LENS],
max_seq_len,
time_major=model.is_time_major())
mask = torch.reshape(mask, [-1])
num_valid = torch.sum(mask)
def reduce_mean_valid(t):
return torch.sum(t[mask]) / num_valid
# non-RNN case: No masking.
else:
mask = None
reduce_mean_valid = torch.mean
prev_action_dist = dist_class(train_batch[SampleBatch.ACTION_DIST_INPUTS],
model)
logp_ratio = torch.exp(
curr_action_dist.logp(train_batch[SampleBatch.ACTIONS]) -
train_batch[SampleBatch.ACTION_LOGP])
action_kl = prev_action_dist.kl(curr_action_dist)
mean_kl_loss = reduce_mean_valid(action_kl)
curr_entropy = curr_action_dist.entropy()
mean_entropy = reduce_mean_valid(curr_entropy)
surrogate_loss = torch.min(
train_batch[Postprocessing.ADVANTAGES] * logp_ratio,
train_batch[Postprocessing.ADVANTAGES] *
torch.clamp(logp_ratio, 1 - policy.config["clip_param"],
1 + policy.config["clip_param"]))
mean_policy_loss = reduce_mean_valid(-surrogate_loss)
# Compute a value function loss.
if policy.config["use_critic"]:
prev_value_fn_out = train_batch[SampleBatch.VF_PREDS]
value_fn_out = model.value_function()
vf_loss1 = torch.pow(
value_fn_out - train_batch[Postprocessing.VALUE_TARGETS], 2.0)
vf_clipped = prev_value_fn_out + torch.clamp(
value_fn_out - prev_value_fn_out, -policy.config["vf_clip_param"],
policy.config["vf_clip_param"])
vf_loss2 = torch.pow(
vf_clipped - train_batch[Postprocessing.VALUE_TARGETS], 2.0)
vf_loss = torch.max(vf_loss1, vf_loss2)
mean_vf_loss = reduce_mean_valid(vf_loss)
# Ignore the value function.
else:
vf_loss = mean_vf_loss = 0.0
total_loss = reduce_mean_valid(-surrogate_loss +
policy.kl_coeff * action_kl +
policy.config["vf_loss_coeff"] * vf_loss -
policy.entropy_coeff * curr_entropy)
# Store values for stats function in model (tower), such that for
# multi-GPU, we do not override them during the parallel loss phase.
model.tower_stats["total_loss"] = total_loss
model.tower_stats["mean_policy_loss"] = mean_policy_loss
model.tower_stats["mean_vf_loss"] = mean_vf_loss
model.tower_stats["vf_explained_var"] = explained_variance(
train_batch[Postprocessing.VALUE_TARGETS], model.value_function())
model.tower_stats["mean_entropy"] = mean_entropy
model.tower_stats["mean_kl_loss"] = mean_kl_loss
return total_loss
