Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[RLlib] Bug fix: DQN goes into negative epsilon values after reaching explora… #6971

Merged
merged 4 commits into from
Jan 31, 2020
Merged

[RLlib] Bug fix: DQN goes into negative epsilon values after reaching explora… #6971

Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
6bca65e
Bug fix: DQN goes into negative epsilon values after reaching explora…
sven1977 Jan 30, 2020
a2386c7
Add `epsilon_initial_eps` to SAC to pass test_nested_spaces.py.
sven1977 Jan 31, 2020
a4f05a7
Merge branch 'master' of https://github.com/ray-project/ray into bug_…
sven1977 Jan 31, 2020
6ef0245
Add `exploration_initial_eps` to QMIX default config.
sven1977 Jan 31, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 13 additions & 7 deletions rllib/agents/dqn/dqn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@
from ray.rllib.agents.dqn.simple_q_policy import SimpleQPolicy
from ray.rllib.optimizers import SyncReplayOptimizer
from ray.rllib.policy.sample_batch import DEFAULT_POLICY_ID
from ray.rllib.utils.schedules import ConstantSchedule, LinearSchedule
from ray.rllib.utils.schedules import ConstantSchedule, PiecewiseSchedule

logger = logging.getLogger(__name__)

Expand Down Expand Up @@ -45,7 +45,9 @@
# Fraction of entire training period over which the exploration rate is
# annealed
"exploration_fraction": 0.1,
# Final value of random action probability
# Initial value of random action probability.
"exploration_initial_eps": 1.0,
# Final value of random action probability.
"exploration_final_eps": 0.02,
# Update the target network every `target_network_update_freq` steps.
"target_network_update_freq": 500,
Expand Down Expand Up @@ -214,11 +216,15 @@ def make_exploration_schedule(config, worker_index):
# local ev should have zero exploration so that eval rollouts
# run properly
return ConstantSchedule(0.0)
return LinearSchedule(
schedule_timesteps=int(
config["exploration_fraction"] * config["schedule_max_timesteps"]),
initial_p=1.0,
final_p=config["exploration_final_eps"])

return PiecewiseSchedule(
endpoints=[
(0, config["exploration_initial_eps"]),
(int(config["exploration_fraction"] *
config["schedule_max_timesteps"]),
config["exploration_final_eps"]),
],
outside_value=config["exploration_final_eps"])


def setup_exploration(trainer):
Expand Down