Views
No views yet
1
2# Defining model
3model = PPO('MlpPolicy', env, n_steps = 512, batch_size = 64, n_epochs = 4, gamma = 0.999, gae_lambda = 0.98, ent_coef = 0.01, verbose=1)
4
5
6# Training
7model.learn(total_timesteps=2000000)
8
9model_name = "ppo-LunarLander-v2"
10model.save(model_name)
11
12# evaluation
13eval_env = Monitor(gym.make("LunarLander-v2"))
14
15# Evaluate the model with 10 evaluation episodes and deterministic=True
16mean_reward, std_reward = evaluate_policy(model, eval_env, n_eval_episodes=10, deterministic=True)
17
18# Print the results
19print(f"mean_reward={mean_reward:.2f} +/- {std_reward}")
20
21# mean_reward=284.85 +/- 18.270698037778157
22...n_steps down to 512total_timestamps up to 2,000,000