def test_LoggingVerbose(self):
ppo_agent = easyagents.tfagents.PpoAgent(
'CartPole-v0',
training_duration=TrainingDurationFast(),
logging=LoggingVerbose())
ppo_agent.train()
return
def test_play_episode(self):
ppo_agent = easyagents.tfagents.PpoAgent(
'CartPole-v0',
training_duration=TrainingDurationSingleEpisode(),
logging=LoggingVerbose())
ppo_agent.train()
TestEasyEnv._step_callback_call_count = 0
(reward, steps) = ppo_agent.play_episode(step_callback)
assert reward > 0
assert steps > 0
assert TestEasyEnv._step_callback_call_count > 0
return
#
# simple easyagents demo using cartpole
from easyagents.tfagents import PpoAgent
from easyagents.config import TrainingDurationSingleEpisode
from easyagents.config import TrainingDurationFast
from easyagents.config import LoggingVerbose
ppo_agent = PpoAgent(gym_env_name='CartPole-v0',
training_duration=TrainingDurationFast(),
logging=LoggingVerbose())
ppo_agent.train()
ppo_agent.plot_average_rewards()
ppo_agent.plot_losses()
ppo_agent.render_episodes_to_mp4()
ppo_agent.render_episodes_to_html(fps=20)
def test_render_episodes_to_str(self):
ppo_agent = PpoAgent( 'CartPole-v0', training_duration=TrainingDurationSingleEpisode(), logging=LoggingVerbose() )
ppo_agent.train()
result = ppo_agent.render_episodes_to_str()
assert isinstance(result,str), "render_episodes_to_str did not return a string"
return
def test_render_episodes(self):
ppo_agent = PpoAgent( 'CartPole-v0', training_duration=TrainingDurationSingleEpisode(), logging=LoggingVerbose() )
ppo_agent.train()
ppo_agent.render_episodes()
return
def test_plot_with_ylim(self):
ppo_agent = PpoAgent( 'CartPole-v0', training_duration=TrainingDurationSingleEpisode(), logging=LoggingVerbose() )
ppo_agent.train()
ppo_agent.plot_losses(ylim=[10,20])
return
def test_plot_no_ylim(self):
ppo_agent = PpoAgent( 'CartPole-v0', training_duration=TrainingDurationSingleEpisode(), logging=LoggingVerbose() )
ppo_agent.train()
ppo_agent.plot_average_rewards()
return