mirror of
https://github.com/vale981/ray
synced 2025-03-05 10:01:43 -05:00
33 lines
978 B
YAML
33 lines
978 B
YAML
# Runs on a single g3.4xl node
|
|
# See https://github.com/ray-project/rl-experiments for results
|
|
dueling-ddqn:
|
|
env:
|
|
grid_search:
|
|
- BreakoutNoFrameskip-v4
|
|
- BeamRiderNoFrameskip-v4
|
|
- QbertNoFrameskip-v4
|
|
- SpaceInvadersNoFrameskip-v4
|
|
run: DQN
|
|
config:
|
|
# Works for both torch and tf.
|
|
framework: tf
|
|
double_q: true
|
|
dueling: true
|
|
num_atoms: 1
|
|
noisy: false
|
|
replay_buffer_config:
|
|
type: MultiAgentReplayBuffer
|
|
capacity: 1000000
|
|
num_steps_sampled_before_learning_starts: 20000
|
|
n_step: 1
|
|
target_network_update_freq: 8000
|
|
lr: .0000625
|
|
adam_epsilon: .00015
|
|
hiddens: [512]
|
|
rollout_fragment_length: 4
|
|
train_batch_size: 32
|
|
exploration_config:
|
|
epsilon_timesteps: 200000
|
|
final_epsilon: 0.01
|
|
num_gpus: 0.2
|
|
min_sample_timesteps_per_iteration: 10000
|