ray/rllib/tuned_examples/regression_tests/cartpole-dqn-param-noise.yaml

18 lines
437 B
YAML
Raw Normal View History

cartpole-dqn-w-param-noise:
env: CartPole-v0
run: DQN
stop:
episode_reward_mean: 150
timesteps_total: 100000
config:
exploration_config:
type: ParameterNoise
random_timesteps: 0
initial_stddev: 1.0
batch_mode: complete_episodes
lr: 0.001
num_workers: 0
model:
fcnet_hiddens: [16]
fcnet_activation: linear