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

19 lines
505 B
YAML

cartpole-dqn-w-param-noise:
env: CartPole-v0
run: DQN
stop:
episode_reward_mean: 150
timesteps_total: 300000
config:
# Works for both torch and tf.
framework: tf
exploration_config:
type: ParameterNoise
random_timesteps: 10000
initial_stddev: 1.0
batch_mode: complete_episodes
lr: 0.0008
num_workers: 0
model:
fcnet_hiddens: [32, 32]
fcnet_activation: tanh