odiaz1066/CartPoleGoal-Lagomorph-seed42

(CleanRL) DQN Agent Playing Lagomorph/CartPoleGoal-v1

This is a trained model of a DQN agent playing a custom environment named Lagomorph/CartPoleGoal-v1. This is a modification of CartPole to include a goal position where the pole should be balanced. The model was trained by using CleanRL.

Get Started

I'll currently update this with the github repo.

Command to reproduce the training

TODO

Hyperparameters

{'batch_size': 128,
 'buffer_size': 10000,
 'capture_video': True,
 'checkpoint': False,
 'checkpoint_frequency': 100000,
 'cuda': True,
 'end_e': 0.05,
 'env_id': 'Lagomorph/CartPoleGoal-v1',
 'exp_name': 'Lagomorph-CartPoleGoal',
 'exploration_fraction': 0.5,
 'gamma': 0.99,
 'hf_entity': 'odiaz1066',
 'initial_steps': 0,
 'learning_rate': 0.00025,
 'learning_starts': 10000,
 'num_envs': 1,
 'resume': True,
 'save_model': True,
 'seed': 42,
 'start_e': 1,
 'target_network_frequency': 500,
 'tau': 1.0,
 'torch_deterministic': True,
 'total_timesteps': 0,
 'track': False,
 'train_frequency': 10,
 'upload_model': True,
 'wandb_entity': None,
 'wandb_project_name': 'lagomorph'}