alperenunlu/LunarLander-v3-dqn

(RLHell) DQN Agent Playing LunarLander-v3

The model was trained by using rlhell.

Command to reproduce the training

curl -OL https://huggingface.co/alperenunlu/LunarLander-v3-dqn/raw/main/dqn.py
curl -OL https://huggingface.co/alperenunlu/LunarLander-v3-dqn/raw/main/pyproject.toml
curl -OL https://huggingface.co/alperenunlu/LunarLander-v3-dqn/raw/main/uv.lock
uv run dqn.py --env-id LunarLander-v3 --n-envs 32 --total-timesteps 2_000_000 --learning-start 100_000 --buffer-size 100_000 --batch-size 1024 --final-exploration 0.01

Hyperparameters

{'batch_size': 1_024,
 'buffer_size': 100_000,
 'env_id': 'LunarLander-v3',
 'eval_episodes': 10,
 'evaluate': True,
 'exp_name': 'dqn',
 'exploration_fraction': 0.5,
 'final_exploration': 0.01,
 'gamma': 0.99,
 'hf_entity': 'alperenunlu',
 'initial_exploration': 1,
 'learning_rate': 0.00025,
 'learning_start': 100_000,
 'log_interval': 100,
 'n_envs': 32,
 'push_model': True,
 'save_times': 10,
 'seed': 0,
 'target_network_update_frequency': 500,
 'tau': 1.0,
 'total_timesteps': 2_000_000,
 'train_frequency': 10,
 'video_capture_frequency': 5}