gryf/coach
1
0
Fork 0
mirror of https://github.com/gryf/coach.git synced 2025-12-17 19:20:19 +01:00
Code Issues Projects Releases Wiki Activity

Cleanup imports.

Browse Source 
Till now, most of the modules were importing all of the module objects
(variables, classes, functions, other imports) into module namespace,
which potentially could (and was) cause of unintentional use of class or
methods, which was indirect imported.

With this patch, all the star imports were substituted with top-level
module, which provides desired class or function.

Besides, all imports where sorted (where possible) in a way pep8[1]
suggests - first are imports from standard library, than goes third
party imports (like numpy, tensorflow etc) and finally coach modules.
All of those sections are separated by one empty line.

[1] https://www.python.org/dev/peps/pep-0008/#imports
This commit is contained in:
Roman Dobosz
2018-04-12 19:46:32 +02:00
parent cafa152382
commit 1b095aeeca
75 changed files with 1169 additions and 1139 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
agents/n_step_q_agent.py
View File

@@ -14,22 +14,21 @@
# limitations under the License.
#
import numpy as np
import scipy.signal
from agents.value_optimization_agent import ValueOptimizationAgent
from agents.policy_optimization_agent import PolicyOptimizationAgent
from logger import logger
from utils import Signal, last_sample
from agents import value_optimization_agent as voa
from agents import policy_optimization_agent as poa
import logger
import utils
# N Step Q Learning Agent - https://arxiv.org/abs/1602.01783
class NStepQAgent(ValueOptimizationAgent, PolicyOptimizationAgent):
class NStepQAgent(voa.ValueOptimizationAgent, poa.PolicyOptimizationAgent):
def __init__(self, env, tuning_parameters, replicated_device=None, thread_id=0):
ValueOptimizationAgent.__init__(self, env, tuning_parameters, replicated_device, thread_id, create_target_network=True)
voa.ValueOptimizationAgent.__init__(self, env, tuning_parameters, replicated_device, thread_id, create_target_network=True)
self.last_gradient_update_step_idx = 0
self.q_values = Signal('Q Values')
self.unclipped_grads = Signal('Grads (unclipped)')
self.value_loss = Signal('Value Loss')
self.q_values = utils.Signal('Q Values')
self.unclipped_grads = utils.Signal('Grads (unclipped)')
self.value_loss = utils.Signal('Value Loss')
self.signals.append(self.q_values)
self.signals.append(self.unclipped_grads)
self.signals.append(self.value_loss)
@@ -57,7 +56,7 @@ class NStepQAgent(ValueOptimizationAgent, PolicyOptimizationAgent):
if game_overs[-1]:
R = 0
else:
R = np.max(self.main_network.target_network.predict(last_sample(next_states)))
R = np.max(self.main_network.target_network.predict(utils.last_sample(next_states)))
for i in reversed(range(num_transitions)):
R = rewards[i] + self.tp.agent.discount * R
@@ -85,4 +84,4 @@ class NStepQAgent(ValueOptimizationAgent, PolicyOptimizationAgent):
else:
logger.create_signal_value('Update Target Network', 0, overwrite=False)
return PolicyOptimizationAgent.train(self)
return poa.PolicyOptimizationAgent.train(self)