weixin_39599654
weixin_39599654
2021-01-07 11:53

error creating PPO agent

layers=[
        dict(type='dense',size=input_size,    activation='relu'),
        dict(type='dense',size=input_size*2,  activation='relu'),
        dict(type='dense',size= output_size*3,activation='relu'),
        dict(type='dense',size=output_size,   activation='relu')
       ]

agent_normal=Agent.create( agent='ppo', environment=environment, network=dict(type='layered',layers=layers), use_beta_distribution='true', memory='minimum', batch_size=10, update_frequency=10, learning_rate=5e-3, multi_step=5, subsampling_fraction=0.91, likelihood_ratio_clipping=0.09, discount=1.0, predict_terminal_values='false', baseline=dict(type='layered',layers=layers), baseline_optimizer=dict(optimizer="adam",learning_rate=5e-3,multi_step=5), state_preprocessing='linear_normalization', reward_preprocessing='null', exploration=exploration, variable_noise=0.0, l2_regularization=0.0, entropy_regularization=0.0
)

该提问来源于开源项目:tensorforce/tensorforce

  • 点赞
  • 回答
  • 收藏
  • 复制链接分享

5条回答