test@... ~/c/o/e/atari (stable) [1]> python train_ppo.py --config atari_ppo.yaml (retro)
Traceback (most recent call last):
File "/home/test/miniconda3/envs/retro/lib/python3.10/site-packages/gymnasium/envs/registration.py", line 802, in make
env = env_creator(**env_spec_kwargs)
TypeError: AtariEnv.__init__() got an unexpected keyword argument 'cfg'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/home/test/code/openrl/examples/atari/train_ppo.py", line 101, in <module>
agent = train()
File "/home/test/code/openrl/examples/atari/train_ppo.py", line 49, in train
env = make(
File "/home/test/miniconda3/envs/retro/lib/python3.10/site-packages/openrl/envs/common/registration.py", line 165, in make
env = AsyncVectorEnv(env_fns, render_mode=render_mode, auto_reset=auto_reset)
File "/home/test/miniconda3/envs/retro/lib/python3.10/site-packages/openrl/envs/vec_env/async_venv.py", line 96, in __init__
dummy_env = env_fns[0]()
File "/home/test/miniconda3/envs/retro/lib/python3.10/site-packages/openrl/envs/common/build_envs.py", line 36, in _make_env
env = make(
File "/home/test/miniconda3/envs/retro/lib/python3.10/site-packages/gymnasium/envs/registration.py", line 814, in make
raise type(e)(
TypeError: AtariEnv.__init__() got an unexpected keyword argument 'cfg' was raised from the environment creator for ALE/Pong-v5 with kwargs ({'game': 'pong', 'obs_type': 'rgb', 'repeat_action_probability': 0.25, 'full_action_space': False, 'frameskip': 4, 'max_num_frames_per_episode': 108000, 'render_mode': None, 'cfg': Namespace(config=[Path_fr(/tmp/tmp4xkjxkro.yaml)], seed=0, encode_state=False, n_block=1, n_embd=64, n_head=1, dec_actor=False, share_actor=False, callbacks=None, sb3_model_path=None, sb3_algo=None, step_difference=1, gail=False, expert_data=None, gail_batch_size=128, dis_input_len=None, gail_loss_target=None, gail_epoch=5, gail_use_action=True, gail_hidden_size=256, gail_layer_num=3, gail_lr=0.0005, data_dir=None, force_rewrite=False, collector_num=1, input_data_dir=None, output_data_dir=None, worker_num=1, sample_interval=1, selfplay_api=Namespace(host='127.0.0.1', port=10086), lazy_load_opponent=True, self_play=False, selfplay_algo='WeightExistEnemy', max_play_num=2000, max_enemy_num=-1, exist_enemy_num=0, random_pos=-1, build_in_pos=-1, use_amp=False, load_optimizer=False, use_joint_action_loss=False, frameskip=None, eval_render=False, terminal='current_terminal', distributed_type='sync', program_type='local', share_temp_dir=None, share_entry_script_path=None, learner_num=1, fetch_num=1, tmux_prefix=None, kill_all=False, namespace='default', mount_path=None, mount_name=None, persistent_volume_claim_name=None, disable_training=False, use_half_actor=False, algorithm_name='ppo', experiment_name='atari_ppo', gpu_usage_type='auto', disable_cuda=False, cuda_deterministic=True, pytorch_threads=1, n_rollout_threads=32, n_eval_rollout_threads=1, n_render_rollout_threads=1, num_env_steps=10000000, user_name='openrl', wandb_entity='openrl-lab', disable_wandb=False, env_name='StarCraft2', scenario_name='default', num_agents=1, num_enemies=1, use_obs_instead_of_state=False, episode_length=128, eval_episode_length=200, max_episode_length=None, separate_policy=False, use_conv1d=False, stacked_frames=1, use_stacked_frames=False, hidden_size=512, layer_N=1, activation_id=1, use_popart=False, dual_clip_ppo=False, dual_clip_coeff=3, use_valuenorm=True, use_feature_normalization=False, use_orthogonal=True, gain=0.01, cnn_layers_params=None, use_maxpool2d=False, rnn_type='gru', rnn_num=1, use_naive_recurrent_policy=False, use_recurrent_policy=False, recurrent_N=1, data_chunk_length=2, use_influence_policy=False, influence_layer_N=1, use_attn=False, attn_N=1, attn_size=64, attn_heads=4, dropout=0.0, use_average_pool=True, use_attn_internal=True, use_cat_self=True, lr=0.00025, tau=0.995, critic_lr=0.00025, opti_eps=1e-05, weight_decay=0, bc_epoch=2, ppo_epoch=4, use_policy_vhead=False, use_clipped_value_loss=True, clip_param=0.1, num_mini_batch=4, mini_batch_size=None, policy_value_loss_coef=0.5, entropy_coef=0.01, value_loss_coef=0.5, use_max_grad_norm=True, max_grad_norm=10.0, use_gae=True, gamma=0.99, gae_lambda=0.95, use_proper_time_limits=False, use_huber_loss=True, use_value_active_masks=True, use_policy_active_masks=True, huber_delta=10.0, use_adv_normalize=True, aux_epoch=5, clone_coef=1.0, use_single_network=False, use_linear_lr_decay=True, save_interval=1, only_eval=False, log_interval=1, log_each_episode=True, use_rich_handler=True, use_eval=False, eval_interval=25, eval_episodes=32, save_gifs=False, use_render=False, render_episodes=5, ifi=0.1, model_dir=None, save_dir=None, init_dir=None, run_dir='./run_results/', use_transmit=False, server_address=None, use_tlaunch=False, actor_num=1, use_reward_normalization=False, buffer_size=5000, popart_update_interval_step=2, use_per=False, per_alpha=0.6, per_beta_start=0.4, per_eps=1e-06, per_nu=0.9, batch_size=32, actor_train_interval_step=2, train_interval_episode=1, train_interval=100, use_same_critic_obs=True, use_global_all_local_state=False, prev_act_inp=False, target_update=10, var=0.5, actor_lr=0.001, auto_alph=False, alpha_value=0.2, alpha_lr=0.0002, use_soft_update=True, hard_update_interval_episode=200, num_random_episodes=5, epsilon_start=1.0, epsilon_finish=0.05, epsilon_anneal_time=5000, use_double_q=True, hypernet_layers=2, mixer_hidden_dim=32, hypernet_hidden_dim=64, target_action_noise_std=0.2, data_path=None, env=Namespace(args={}), model_path=None, use_share_model=True, reward_class=Namespace(id=None, args={}), vec_info_class=Namespace(id='EPS_RewardInfo', args={}), eval_metrics=[], disable_update_enemy=False, least_win_rate=0.5, recent_list_max_len=100, latest_weight=0.5, newest_pos=1, newest_weight=0.5)})
🐛 Bug
Having problem while running Atari example code on both stable and main branch
It seems due to the cfg didn't correctly pass to gymnasium make
To Reproduce
Relevant log output / Error message
System Info
Checklist