Skip to content
Permalink
Branch: master
Find file Copy path
Find file Copy path
1 contributor

Users who have contributed to this file

37 lines (36 sloc) 1.03 KB
# Our implementation of SAC can reach 9k reward in 400k timesteps
halfcheetah_sac:
env: HalfCheetah-v3
run: SAC
stop:
episode_reward_mean: 9000
config:
horizon: 1000
soft_horizon: False
Q_model:
hidden_activation: relu
hidden_layer_sizes: [256, 256]
policy_model:
hidden_activation: relu
hidden_layer_sizes: [256, 256]
tau: 0.005
target_entropy: auto
no_done_at_end: True
n_step: 1
sample_batch_size: 1
prioritized_replay: False
train_batch_size: 256
target_network_update_freq: 1
timesteps_per_iteration: 1000
learning_starts: 10000
exploration_enabled: True
optimization:
actor_learning_rate: 0.0003
critic_learning_rate: 0.0003
entropy_learning_rate: 0.0003
num_workers: 0
num_gpus: 0
clip_actions: False
normalize_actions: True
evaluation_interval: 1
metrics_smoothing_episodes: 5
You can’t perform that action at this time.