from easydict import EasyDict
import ding.envs.gym_env

cfg = dict(
    exp_name='PongNoFrameskip-v4-PPOOffPolicy',
    env=dict(
        collector_env_num=8,
        evaluator_env_num=8,
        n_evaluator_episode=8,
        stop_value=30,
        env_id='PongNoFrameskip-v4',
        frame_stack=4,
        env_wrapper='atari_default',
    ),
    policy=dict(
        cuda=True,
        recompute_adv=True,
        action_space='discrete',
        model=dict(
            obs_shape=[4, 84, 84],
            action_shape=6,
            action_space='discrete',
            encoder_hidden_size_list=[64, 64, 128],
            actor_head_hidden_size=128,
            critic_head_hidden_size=128,
        ),
        learn=dict(
            update_per_collect=10,
            batch_size=320,
            learning_rate=3e-4,
            value_weight=0.5,
            entropy_weight=0.001,
            clip_ratio=0.2,
            adv_norm=True,
            # value_norm=True,
            ignore_done=False,
            grad_clip_type='clip_norm',
            grad_clip_value=0.5,
        ),
        collect=dict(
            n_sample=3200,
            unroll_len=1,
            discount_factor=0.99,
            gae_lambda=0.95,
        ),
    ),
    wandb_logger=dict(
        gradient_logger=True, video_logger=True, plot_logger=True, action_logger=True, return_logger=False
    ),
)

cfg = EasyDict(cfg)

env = ding.envs.gym_env.env