Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

经验回放 #53

Open
shuzi opened this issue Mar 15, 2023 · 1 comment
Open

经验回放 #53

shuzi opened this issue Mar 15, 2023 · 1 comment

Comments

@shuzi
Copy link

shuzi commented Mar 15, 2023

3_DQN_1.pdf 里面 importance sampling 需要scaling learning rate , 请问为什么重要性高的经验要用比较低的learning rate

@DCLe-DA
Copy link

DCLe-DA commented Mar 30, 2023

3_DQN_1.pdf 里面 importance sampling 需要scaling learning rate , 请问为什么重要性高的经验要用比较低的learning rate

您好我是这么理解的,按照公式的话分母上的数越大这个数越小;按照实际情况的话老师提到了“某类state出现的频率比较低,比如超级玛丽中的BOSS关卡”,那么我的理解是你可参考的场景(state)转移(transmission)也会相应的变少,不利于你去做参数的调整。假如你将学习率设置的太大,那么在曲线上的表示就是你很容易跳过最优点。而出现频率较多的states因为参考很多进而在误差曲线占据的范围也多,较大的学习率可以加快搜索。

# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants