mirror of
https://github.com/vale981/ray
synced 2025-03-06 02:21:39 -05:00
32 lines
919 B
YAML
32 lines
919 B
YAML
stateless-cartpole-r2d2:
|
|
env: ray.rllib.examples.env.stateless_cartpole.StatelessCartPole
|
|
run: R2D2
|
|
stop:
|
|
episode_reward_mean: 150
|
|
timesteps_total: 1000000
|
|
config:
|
|
# Works for both torch and tf.
|
|
framework: tf
|
|
num_workers: 0
|
|
# R2D2 settings.
|
|
replay_buffer_config:
|
|
type: MultiAgentReplayBuffer
|
|
storage_unit: sequences
|
|
replay_burn_in: 20
|
|
zero_init_states: true
|
|
#dueling: false
|
|
lr: 0.0005
|
|
# Give some more time to explore.
|
|
exploration_config:
|
|
epsilon_timesteps: 50000
|
|
# Wrap with an LSTM and use a very simple base-model.
|
|
model:
|
|
fcnet_hiddens: [64]
|
|
fcnet_activation: linear
|
|
use_lstm: true
|
|
lstm_cell_size: 64
|
|
max_seq_len: 20
|
|
|
|
# Fake 2 GPUs.
|
|
num_gpus: 2
|
|
_fake_gpus: true
|