0

I am running some simulations using PPO and A2C algorithms from Stablebaselines3 with openai-gym. I know that i can customize all of them, but i was wondering which are the default parameters. Specially the network structure and learning rate...

Does anyone know these values or have any clue where I can find them?

Thanks in advance, Samuel

I haven't found it at the stable baselines docs..

Samuel
  • 3
  • 2
  • you may also find inspiring some pre-tuned hypeparameter sets from rl-baselines3-zoo package from the same developers. Check: https://github.com/DLR-RM/rl-baselines3-zoo/tree/master/hyperparams – gehirndienst Feb 23 '23 at 12:09

1 Answers1

0

For A2C, the values you're looking for are located in the docs here: https://stable-baselines3.readthedocs.io/en/master/modules/a2c.html#stable_baselines3.a2c.A2C

class stable_baselines3.a2c.A2C(policy, env, learning_rate=0.0007, n_steps=5, gamma=0.99, gae_lambda=1.0, ent_coef=0.0, vf_coef=0.5, max_grad_norm=0.5, rms_prop_eps=1e-05, use_rms_prop=True, use_sde=False, sde_sample_freq=-1, normalize_advantage=False, tensorboard_log=None, policy_kwargs=None, verbose=0, seed=None, device='auto', _init_setup_model=True)

For PPO, the values are located here: https://stable-baselines3.readthedocs.io/en/master/modules/ppo.html#stable_baselines3.ppo.PPO

class stable_baselines3.ppo.PPO(policy, env, learning_rate=0.0003, n_steps=2048, batch_size=64, n_epochs=10, gamma=0.99, gae_lambda=0.95, clip_range=0.2, clip_range_vf=None, normalize_advantage=True, ent_coef=0.0, vf_coef=0.5, max_grad_norm=0.5, use_sde=False, sde_sample_freq=-1, target_kl=None, tensorboard_log=None, policy_kwargs=None, verbose=0, seed=None, device='auto', _init_setup_model=True)

The docs contain brief explanations of each of the parameters, but if something is unclear, feel free to ask.

If these weren't the parameters you were looking for, please let me know.

fraz
  • 26
  • 4