minetester-treechop_shaped-v0 deep-reinforcement-learning reinforcement-learning custom-implementation

#PPO Agent Playing minetester-treechop_shaped-v0

This is a trained model of a PPO agent playing minetester-treechop_shaped-v0. The model was trained by using minetest-baselines.

Command to reproduce the training

python -m minetest_baselines.train --algo ppo --exp-name ppo_fixed_dtime_seed1 --seed 1 --capture-video --track --wandb-entity rkla --save-model --upload-model --hf-entity rkla --total-timesteps 1000 --num-envs 2

Hyperparameters

{'anneal_lr': True,
 'batch_size': 256,
 'capture_video': True,
 'clip_coef': 0.1,
 'cuda': True,
 'ent_coef': 0.01,
 'env_id': 'minetester-treechop_shaped-v0',
 'exp_name': 'ppo_fixed_dtime_seed1',
 'gae_lambda': 0.95,
 'gamma': 0.99,
 'hf_entity': 'rkla',
 'learning_rate': 0.00025,
 'max_grad_norm': 0.5,
 'minibatch_size': 64,
 'norm_adv': True,
 'num_envs': 2,
 'num_minibatches': 4,
 'num_steps': 128,
 'num_updates': 3,
 'save_model': True,
 'seed': 1,
 'target_kl': None,
 'torch_deterministic': True,
 'total_timesteps': 1000,
 'track': True,
 'update_epochs': 4,
 'upload_model': True,
 'vf_coef': 0.5,
 'video_frequency': 100,
 'wandb_entity': 'rkla',
 'wandb_project_name': 'minetest-baselines'}