Jianing issues

Results 4 issues of


                                            Jianing

I want to train an agent with mujoco env, but don't know how to set the hyperparameter, is there and demo.py for mujoco env?

help_wanted

您好，我直接使用demo_A2C_PPO.py训练pendulum环境下的A2C算法无法收敛，可能算法实现上有问题。AgentDiscreteA2C算法仅继承了AgentDiscretePPO，并未实现自己的update_net函数

bug

第235行部分self.cri.state_std[:] = self.cri.state_std，这样的话cri.state_std是永远不变的

bug

train_args只有13b的训练配置