-
老师好,6.3 节对决网络(Dueling Network)好像没有对 Dueling DQN “为什么要把 Q 值函数拆开” 的解释,所以我最开始看完了这一节后依然有点困惑,所以希望可以加一些这部分的解释。(当然如果是我遗漏了这一部分的话,那非常抱歉 😂)
我目前对 Dueling DQN 的粗浅的理解是,它拆 Q 值函数是为了把状态和动作分开考虑,从而能够判断 Q 值高到底是因为状态好所…
-
当使用demo_DQN_Dueling_Double_DQN 训练结束的的pt文件无法作为测试时的权重文件 ,是否需要将保存pt文件
由torch.save(actor, actor_path)
更改为torch.save(actor.state_dict(), actor_path)
-
line 56
` advantage = tl.layers.ElementwiseLambda(lambda x,y: x-y)([avalue,mean]) #a - avg(a)`
the variable "advantage" is not used, anything wrong?
-
Originally reported by: **Kneth (Bitbucket: [kennyt](https://bitbucket.org/kennyt), GitHub: [kennyt](https://github.com/kennyt))**
---
As a hunter, if you're dueling a group member, your pet won't a…
-
### Ability name
Spell Steal, Duel
### Description
If Rubick steals Duel and duels a target that is already dueled, the first duel to end will end both duels. This seems like unintentional behavior…
-
### Ability name
Duel
### Description
When duel is cast on a target that is already dueling, which ever instance of duel finishes first decides when the others finish.
The dota 2 wiki page is th…
-
Hello, I would like to ask where your code uses D3QN. I was confused when I saw TD3 in the train.py code, and would like to ask you what happened?Additionally, I am also confused about the use of duel…
-
It would be cool if some kind of flag or mark spawned when dueling so you know the radius and where the middle is,
i was thinking maby 3 fences on top of eachother and a whool as "flag"
something l…
-
Please, I have the following issue while trying to run main_dueling_ddqn.py
![essue](https://user-images.githubusercontent.com/58139310/144785226-5b7ee8b4-a0a6-40e5-aa13-45bca66595fe.PNG)
-
Smuggler
Camouflage Ally no longer works on a target, if you are dueling that target.
Cekis updated
3 years ago