def auto_td3_benchmarks():
"""Run experiments for TD3 benchmarking."""
td3_env_ids = [
env_id for env_id in MuJoCo1M_ENV_SET if env_id != 'Reacher-v2'
]
iterate_experiments(td3_garage_tf, td3_env_ids)
def auto_vpg_benchmarks():
"""Run experiments for VPG benchmarking."""
iterate_experiments(vpg_garage_pytorch,
MuJoCo1M_ENV_SET,
snapshot_config={'snapshot_mode': 'none'})
iterate_experiments(vpg_garage_tf,
MuJoCo1M_ENV_SET,
snapshot_config={'snapshot_mode': 'none'})
def auto_trpo_benchmarks(snapshot_config):
"""Run experiments for TRPO benchmarking."""
iterate_experiments(trpo_garage_pytorch,
MuJoCo1M_ENV_SET,
snapshot_config=snapshot_config)
iterate_experiments(trpo_garage_tf,
MuJoCo1M_ENV_SET,
snapshot_config=snapshot_config)
def auto_td3_benchmarks():
"""Run experiments for TD3 benchmarking."""
td3_env_ids = [
env_id for env_id in MuJoCo1M_ENV_SET if env_id != 'Reacher-v2'
]
iterate_experiments(td3_garage_tf,
td3_env_ids,
snapshot_config={'snapshot_mode': 'none'})
def continuous_mlp_baseline_tf_ppo_benchmarks():
"""Run benchmarking experiments for Continuous MLP Baseline on TF-PPO."""
iterate_experiments(continuous_mlp_baseline,
MuJoCo1M_ENV_SET,
seeds=_seeds)
def auto_vpg_benchmarks():
"""Run experiments for VPG benchmarking."""
iterate_experiments(vpg_garage_pytorch, MuJoCo1M_ENV_SET)
iterate_experiments(vpg_garage_tf, MuJoCo1M_ENV_SET)
def auto_trpo_benchmarks():
"""Run experiments for TRPO benchmarking."""
iterate_experiments(trpo_garage_pytorch, MuJoCo1M_ENV_SET)
iterate_experiments(trpo_garage_tf, MuJoCo1M_ENV_SET)
def auto_ddpg_benchmarks():
"""Run experiments for DDPG benchmarking."""
iterate_experiments(ddpg_garage_tf, MuJoCo1M_ENV_SET)
def continuous_mlp_q_function_tf_ddpg_benchmarks():
"""Run benchmarking experiments for Continuous MLP QFunction on TF-DDPG."""
iterate_experiments(continuous_mlp_q_function,
MuJoCo1M_ENV_SET,
seeds=_seeds)
def continuous_mlp_policy_tf_ddpg_benchmarks():
"""Run benchmarking experiments for Continuous MLP Policy on TF-DDPG."""
seeds = random.sample(range(100), 5)
iterate_experiments(continuous_mlp_policy, MuJoCo1M_ENV_SET, seeds=seeds)
def categorical_mlp_policy_tf_ppo_benchmarks():
"""Run benchmarking experiments for Categorical MLP Policy on TF-PPO."""
iterate_experiments(categorical_mlp_policy, STATE_ENV_SET, seeds=_seeds)
def categorical_cnn_policy_tf_ppo_benchmarks():
"""Run benchmarking experiments for Categorical CNN Policy on TF-PPO."""
iterate_experiments(categorical_cnn_policy, PIXEL_ENV_SET, seeds=_seeds)
def her_benchmarks():
"""Run experiments for HER benchmarking."""
iterate_experiments(her_garage_tf, Fetch1M_ENV_SET)
def gaussian_cnn_baseline_tf_ppo_benchmarks():
"""Run benchmarking experiments for Gaussian CNN Baseline on TF-PPO."""
iterate_experiments(gaussian_cnn_baseline, PIXEL_ENV_SET, seeds=_seeds)
def gaussian_mlp_policy_tf_ppo_benchmarks():
"""Run benchmarking experiments for Gaussian MLP Policy on TF-PPO."""
iterate_experiments(gaussian_mlp_policy, MuJoCo1M_ENV_SET, seeds=_seeds)
def gaussian_mlp_baseline_tf_ppo_benchmarks():
"""Run benchmarking experiments for Gaussian MLP Baseline on TF-PPO."""
iterate_experiments(gaussian_mlp_baseline, MuJoCo1M_ENV_SET, seeds=_seeds)
def auto_ddpg_benchmarks(snapshot_config):
"""Run experiments for DDPG benchmarking."""
iterate_experiments(ddpg_garage_tf,
MuJoCo1M_ENV_SET,
snapshot_config=snapshot_config)