英文

A2C Agent玩BreakoutNoFrameskip-v4

这是一个经过训练的A2C代理模型,用于玩BreakoutNoFrameskip-v4,使用了 stable-baselines3 library RL Zoo

RL Zoo是一个用于稳定基线3加强学习代理的训练框架,包含了超参数优化和预训练代理。

使用(使用SB3 RL Zoo)

RL Zoo: https://github.com/DLR-RM/rl-baselines3-zoo SB3: https://github.com/DLR-RM/stable-baselines3 SB3 Contrib: https://github.com/Stable-Baselines-Team/stable-baselines3-contrib

# Download model and save it into the logs/ folder
python -m rl_zoo3.load_from_hub --algo a2c --env BreakoutNoFrameskip-v4 -orga sb3 -f logs/
python enjoy.py --algo a2c --env BreakoutNoFrameskip-v4  -f logs/

训练(使用RL Zoo)

python train.py --algo a2c --env BreakoutNoFrameskip-v4 -f logs/
# Upload the model and generate video (when possible)
python -m rl_zoo3.push_to_hub --algo a2c --env BreakoutNoFrameskip-v4 -f logs/ -orga sb3

超参数

OrderedDict([('ent_coef', 0.01),
             ('env_wrapper',
              ['stable_baselines3.common.atari_wrappers.AtariWrapper']),
             ('frame_stack', 4),
             ('n_envs', 16),
             ('n_timesteps', 10000000.0),
             ('policy', 'CnnPolicy'),
             ('policy_kwargs',
              'dict(optimizer_class=RMSpropTFLike, '
              'optimizer_kwargs=dict(eps=1e-5))'),
             ('vf_coef', 0.25),
             ('normalize', False)])