new_height=84,
to_gray=True,
noop_action=[1, 0, 0, 0],
start_action=[0, 1, 0, 0],
clip_rewards=True)
test_env = AtariWrapper(env_name,
action_repeat=4,
obs_stack=4,
new_width=84,
new_height=84,
to_gray=True,
start_action=[0, 1, 0, 0])
agent = DeepQ(env=env,
use_double=True,
model=DeepQModel(nature_arch=True, dueling=False),
optimizer=RMSProp(7e-4, decay=0.99, epsilon=0.1),
policy=EGreedyPolicy(1.0, 0.1, 4000000),
targetfreq=10000)
trainer = ReplayTrainer(env=env,
agent=agent,
maxsteps=80000000,
replay=backPropagationReplay,
logdir='tmp/%s/moving_average_bias' % env_name,
logfreq=900,
test_env=test_env,
test_render=False)
trainer.train()
new_height=84,
to_gray=True,
start_action=[0, 1, 0, 0, 0, 0])
agent = DeepQ(env,
use_double=False,
model=DeepQModel(nature_arch=False, dueling=False),
optimizer=RMSProp(7e-4, decay=0.99, epsilon=0.1))
threads = []
for i, eps in enumerate([0.01, 0.01, 0.01, 0.1, 0.1, 0.1, 0.5, 0.5] * 2):
threads.append(
DeepQ(env.copy(),
model=DeepQModel(nature_arch=False, dueling=False),
optimizer=agent.opt,
trainable_weights=agent.weights,
target_net=agent.target_net,
target_weights=agent.target_weights,
use_double=False,
targetfreq=10000,
policy=EGreedyPolicy(1.0, eps, 4000000),
name='thread%s' % i))
trainer = AsyncTrainer(agent,
threads,
maxsteps=80000000,
batch_size=5,
logdir='/tmp/rf/AsyncDeepQ/%s' % env_name,
logfreq=240,
test_env=test_env)
trainer.train()
from reinforceflow.core.optimizer import Adam
from reinforceflow.trainers.async_trainer import AsyncTrainer
reinforceflow.set_random_seed(555)
env_name = 'CartPole-v0'
agent = DeepQ(Vectorize(env_name),
model=FullyConnected(),
optimizer=Adam(3e-5))
threads = []
for i, eps in enumerate([0.8, 0.4] * 2):
threads.append(
DeepQ(Vectorize(env_name),
model=FullyConnected(),
optimizer=agent.opt,
trainable_weights=agent.weights,
target_net=agent.target_net,
target_weights=agent.target_weights,
targetfreq=10000,
policy=EGreedyPolicy(1, eps, 100000),
name='Thread%s' % i))
trainer = AsyncTrainer(agent,
threads,
maxsteps=100000,
batch_size=20,
logdir='/tmp/rf/AsyncDeepQ/%s' % env_name,
logfreq=10)
trainer.train()