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

network_imporvements branch merge

Browse Source
This commit is contained in:
Shadi Endrawis
2018-10-02 13:41:46 +03:00
parent 72ea933384
commit 51726a5b80
110 changed files with 1639 additions and 1161 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
rl_coach/presets/Atari_Dueling_DDQN_with_PER_OpenAI.py
View File

@@ -1,22 +1,13 @@
from rl_coach.agents.ddqn_agent import DDQNAgentParameters
from rl_coach.architectures.tensorflow_components.heads.dueling_q_head import DuelingQHeadParameters
from rl_coach.base_parameters import VisualizationParameters, MiddlewareScheme, PresetValidationParameters
from rl_coach.core_types import EnvironmentSteps, RunPhase
from rl_coach.environments.environment import MaxDumpMethod, SelectedPhaseOnlyDumpMethod, SingleLevelSelection
from rl_coach.environments.gym_environment import Atari, atari_deterministic_v4
from rl_coach.core_types import EnvironmentSteps
from rl_coach.environments.environment import SingleLevelSelection
from rl_coach.environments.gym_environment import Atari, atari_deterministic_v4, atari_schedule
from rl_coach.graph_managers.basic_rl_graph_manager import BasicRLGraphManager
from rl_coach.graph_managers.graph_manager import ScheduleParameters
from rl_coach.memories.non_episodic.prioritized_experience_replay import PrioritizedExperienceReplayParameters
from rl_coach.schedules import LinearSchedule, PieceWiseSchedule, ConstantSchedule
####################
# Graph Scheduling #
####################
schedule_params = ScheduleParameters()
schedule_params.improve_steps = EnvironmentSteps(50000000)
schedule_params.steps_between_evaluation_periods = EnvironmentSteps(250000)
schedule_params.evaluation_steps = EnvironmentSteps(135000)
schedule_params.heatup_steps = EnvironmentSteps(50000)
#########
# Agent #
@@ -38,12 +29,7 @@ agent_params.memory.beta = LinearSchedule(0.4, 1, 12500000) # 12.5M training it
###############
# Environment #
###############
env_params = Atari()
env_params.level = SingleLevelSelection(atari_deterministic_v4)
vis_params = VisualizationParameters()
vis_params.video_dump_methods = [SelectedPhaseOnlyDumpMethod(RunPhase.TEST), MaxDumpMethod()]
vis_params.dump_mp4 = False
env_params = Atari(level=SingleLevelSelection(atari_deterministic_v4))
########
# Test #
@@ -52,5 +38,5 @@ preset_validation_params = PresetValidationParameters()
preset_validation_params.trace_test_levels = ['breakout', 'pong', 'space_invaders']
graph_manager = BasicRLGraphManager(agent_params=agent_params, env_params=env_params,
schedule_params=schedule_params, vis_params=vis_params,
schedule_params=atari_schedule, vis_params=VisualizationParameters(),
preset_validation_params=preset_validation_params)