Moki zzZ

某格空着的话，仍然会出问题。例如： ``` a | b | c ---|---|--- a | | ``` 第二行后两格应该是空的，但实际效果是： a | b | c -- | -- | -- a | \|

 teadocs版本号v0.4.0，是版本的原因吗？

解决办法还是有的，就是在空白格中填充``。

感谢作者+1

Mine are better than you. I got r1 30, map 15, and train from scratch. (batch_size=96, lr=2e-4, Duke->Market)

try disable all your browser extensions, or open the page in a clean browser.

Already fixed in the newest code: https://github.com/seungeunrho/minimalRL/blob/c8efed8481e3cd40e9739cfde220a55522555b57/actor_critic.py#L75 https://github.com/seungeunrho/minimalRL/blob/c8efed8481e3cd40e9739cfde220a55522555b57/actor_critic.py#L81 But have not been fixed in other algorithms like `a3c.py`: https://github.com/seungeunrho/minimalRL/blob/c8efed8481e3cd40e9739cfde220a55522555b57/a3c.py#L48 https://github.com/seungeunrho/minimalRL/blob/c8efed8481e3cd40e9739cfde220a55522555b57/a3c.py#L55

It seems like the long time cost at lines 93-102 of `dqn.py` is due to the gameplay process using the well-trained Qnet. https://github.com/seungeunrho/minimalRL/blob/c8efed8481e3cd40e9739cfde220a55522555b57/dqn.py#L93-L102 This longer duration is not a problem...