"""
Qmix: Monotonic value function factorisation for deep multi-agent reinforcement learning
Paper link:
http://proceedings.mlr.press/v80/rashid18a/rashid18a.pdf
Implementation: TensorFlow 2.X
"""
from argparse import Namespace
from operator import itemgetter
from xuance.common import List
from xuance.tensorflow import tf, tk, Module
from xuance.tensorflow.learners import LearnerMAS
[docs]
class QMIX_Learner(LearnerMAS):
def __init__(self,
config: Namespace,
model_keys: List[str],
agent_keys: List[str],
policy: Module,
callback):
super(QMIX_Learner, self).__init__(config, model_keys, agent_keys, policy, callback)
self.build_optimizer()
self.gamma = config.gamma
self.sync_frequency = config.sync_frequency
self.n_actions = {k: self.policy.action_space[k].n for k in self.model_keys}
self.mse_loss = tk.losses.MeanSquaredError()
[docs]
def build_optimizer(self):
if ("macOS" in self.os_name) and ("arm" in self.os_name): # For macOS with Apple's M-series chips.
self.optimizer = tk.optimizers.legacy.Adam(self.config.learning_rate)
else:
self.optimizer = tk.optimizers.Adam(self.config.learning_rate)
@tf.function
def forward_fn(self, bs, state, obs, actions, rewards_tot, state_next, obs_next, terminals_tot,
agent_mask, avail_actions, avail_actions_next, IDs):
with tf.GradientTape() as tape:
_, _, q_eval = self.policy(observation=obs, agent_ids=IDs, avail_actions=avail_actions)
_, q_next = self.policy.Qtarget(observation=obs_next, agent_ids=IDs)
q_eval_a, q_next_a = {}, {}
for key in self.model_keys:
q_eval_a[key] = tf.reshape(tf.gather(q_eval[key], tf.cast(actions[key][:, None], dtype=tf.int32),
axis=-1, batch_dims=-1), [bs])
if self.use_actions_mask:
q_next[key][avail_actions_next[key] == 0] = -1e10
if self.config.double_q:
_, act_next, _ = self.policy(observation=obs_next, agent_ids=IDs,
avail_actions=avail_actions, agent_key=key)
q_next_a[key] = tf.reshape(tf.gather(q_next[key], act_next[key][:, None],
axis=-1, batch_dims=-1), [bs])
else:
q_next_a[key] = q_next[key].max(dim=-1, keepdim=True).values.reshape(bs)
q_eval_a[key] *= agent_mask[key]
q_next_a[key] *= agent_mask[key]
q_tot_eval = self.policy.Q_tot(q_eval_a, state)
q_tot_next = self.policy.Qtarget_tot(q_next_a, state_next)
q_tot_target = rewards_tot + (1 - terminals_tot) * self.gamma * q_tot_next
q_tot_target = tf.reshape(q_tot_target, [-1])
q_tot_eval = tf.reshape(q_tot_eval, [-1])
# calculate the loss function
loss = self.mse_loss(tf.stop_gradient(q_tot_target), q_tot_eval)
gradients = tape.gradient(loss, self.policy.parameters_model)
if self.use_grad_clip:
gradients, _ = tf.clip_by_global_norm(gradients, clip_norm=self.grad_clip_norm)
self.optimizer.apply_gradients(zip(gradients, self.policy.parameters_model))
else:
self.optimizer.apply_gradients(zip(gradients, self.policy.parameters_model))
return loss, tf.math.reduce_mean(q_tot_eval)
@tf.function
def learn(self, *inputs):
if self.distributed_training:
loss, predictQ = self.policy.mirrored_strategy.run(self.forward_fn, args=inputs)
return (self.policy.mirrored_strategy.reduce(tf.distribute.ReduceOp.SUM, loss, axis=None),
self.policy.mirrored_strategy.reduce(tf.distribute.ReduceOp.SUM, predictQ, axis=None))
else:
return self.forward_fn(*inputs)
[docs]
def update(self, sample):
self.iterations += 1
# prepare training data
sample_Tensor = self.build_training_data(sample=sample,
use_parameter_sharing=self.use_parameter_sharing,
use_actions_mask=self.use_actions_mask,
use_global_state=True)
batch_size = sample_Tensor['batch_size']
state = sample_Tensor['state']
state_next = sample_Tensor['state_next']
obs = sample_Tensor['obs']
actions = sample_Tensor['actions']
obs_next = sample_Tensor['obs_next']
rewards = sample_Tensor['rewards']
terminals = sample_Tensor['terminals']
agent_mask = sample_Tensor['agent_mask']
avail_actions = sample_Tensor['avail_actions']
avail_actions_next = sample_Tensor['avail_actions_next']
IDs = sample_Tensor['agent_ids']
if self.use_parameter_sharing:
key = self.model_keys[0]
bs = batch_size * self.n_agents
rewards_tot = tf.reshape(tf.reduce_mean(rewards[key], axis=1), [batch_size, 1])
terminals_tot = tf.reshape(tf.reduce_prod(terminals[key], axis=1), [batch_size, 1])
else:
bs = batch_size
rewards_tot = tf.reduce_mean(tf.stack(itemgetter(*self.agent_keys)(rewards), axis=1),
axis=-1, keepdims=True)
terminals_tot = tf.reduce_prod(tf.stack(itemgetter(*self.agent_keys)(terminals), axis=1),
axis=1, keepdims=True)
info = self.callback.on_update_start(self.iterations, method="update", policy=self.policy,
sample_Tensor=sample_Tensor, bs=bs,
rewards_tot=rewards_tot, terminals_tot=terminals_tot)
loss, predictQ = self.learn(bs, state, obs, actions, rewards_tot, state_next, obs_next, terminals_tot,
agent_mask, avail_actions, avail_actions_next, IDs)
info.update({
"loss_Q": loss.numpy(),
"predictQ": predictQ.numpy()
})
if self.iterations % self.sync_frequency == 0:
self.policy.copy_target()
info.update(self.callback.on_update_end(self.iterations, method="update", policy=self.policy, info=info,
predictQ=predictQ, loss=loss))
return info