Add code

README.md

@@ -24,5 +24,59 @@ model-index:
   This is a trained model of a **DQN** agent playing **LunarLander-v2** using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
   
   ## Usage (with Stable-baselines3)
-  TODO: Add your code
+
+```python
+from stable_baselines3 import DQN
+from stable_baselines3.common.env_util import make_vec_env
+from stable_baselines3.common.callbacks import EvalCallback
+
+# Create the environment
+env_id = "LunarLander-v2"
+n_envs = 1
+env = make_vec_env(env_id, n_envs=n_envs)
+
+# Create the evaluation envs
+eval_envs = make_vec_env(env_id, n_envs=5)
+
+# Adjust evaluation interval depending on the number of envs
+eval_freq = int(1e5)
+eval_freq = max(eval_freq // n_envs, 1)
+
+# Create evaluation callback to save best model
+# and monitor agent performance
+eval_callback = EvalCallback(
+    eval_envs,
+    best_model_save_path="./logs/",
+    eval_freq=eval_freq,
+    n_eval_episodes=10,
+)
+
+# Instantiate the agent
+# Hyperparameters from https://github.com/DLR-RM/rl-baselines3-zoo
+model = DQN(
+    "MlpPolicy",
+    env,
+    learning_starts=0,
+    batch_size=128,
+    buffer_size=50000,
+    learning_rate=1e-3,
+    target_update_interval=250,
+    train_freq=4,
+    gradient_steps=-1,
+    # Explore for 20_000 timesteps
+    exploration_fraction=0.04,
+    exploration_final_eps=0.1,
+    policy_kwargs=dict(net_arch=[256, 256]),
+    verbose=1,
+)
+
+# Train the agent (you can kill it before using ctrl+c)
+try:
+    model.learn(total_timesteps=int(5e5), callback=eval_callback)
+except KeyboardInterrupt:
+    pass
+
+# Load best model
+model = DQN.load("logs/best_model.zip")
+```