gryf/coach
1
0
Fork 0
mirror of https://github.com/gryf/coach.git synced 2025-12-18 11:40:18 +01:00
Code Issues Projects Releases Wiki Activity

adding dueling support for rainbow dqn (now only missing n-step)

Browse Source
This commit is contained in:
Gal Leibovich
2018-08-30 18:15:59 +03:00
parent d2623c0eee
commit ea294de7fd
3 changed files with 40 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
rl_coach/agents/rainbow_dqn_agent.py
View File

@@ -18,25 +18,20 @@ from typing import Union
import numpy as np
from rl_coach.agents.categorical_dqn_agent import CategoricalDQNNetworkParameters, CategoricalDQNAlgorithmParameters, \
from rl_coach.agents.categorical_dqn_agent import CategoricalDQNAlgorithmParameters, \
CategoricalDQNAgent, CategoricalDQNAgentParameters
from rl_coach.agents.dqn_agent import DQNNetworkParameters, DQNAlgorithmParameters
from rl_coach.agents.value_optimization_agent import ValueOptimizationAgent
from rl_coach.architectures.tensorflow_components.heads.categorical_q_head import CategoricalQHeadParameters
from rl_coach.base_parameters import AgentParameters
from rl_coach.agents.dqn_agent import DQNNetworkParameters
from rl_coach.architectures.tensorflow_components.heads.rainbow_q_head import RainbowQHeadParameters
from rl_coach.exploration_policies.parameter_noise import ParameterNoiseParameters
from rl_coach.memories.non_episodic.experience_replay import ExperienceReplayParameters
from rl_coach.memories.non_episodic.prioritized_experience_replay import PrioritizedExperienceReplayParameters, \
PrioritizedExperienceReplay
from rl_coach.schedules import LinearSchedule
from rl_coach.core_types import StateType
from rl_coach.exploration_policies.e_greedy import EGreedyParameters
class RainbowDQNNetworkParameters(CategoricalDQNNetworkParameters):
class RainbowDQNNetworkParameters(DQNNetworkParameters):
def __init__(self):
super().__init__()
self.heads_parameters = [RainbowQHeadParameters()]
class RainbowDQNAlgorithmParameters(CategoricalDQNAlgorithmParameters):
@@ -68,10 +63,11 @@ class RainbowDQNAgentParameters(CategoricalDQNAgentParameters):
# 2. C51
# 3. Prioritized ER
# 4. DDQN
# 5. Dueling DQN
#
# still missing:
# 1. N-Step
# 2. Dueling DQN
class RainbowDQNAgent(CategoricalDQNAgent):
def __init__(self, agent_parameters, parent: Union['LevelManager', 'CompositeAgent']=None):
super().__init__(agent_parameters, parent)