Comments (9)
@FYNIXqwq 没错 更改牌库是有效果的
from douzero.
@daochenzha 目前不洗牌模式训练有一个很大的难题,当发牌风格改为不洗牌的类型(牌型整齐,而且bomb_count很高)的时候,训练的初期loss会非常高,数量级可以从10^2一路升到10^4,请教一下在这种情况下应该如何调节参数?(batch_size个人尝试过16和32,learning_rate从10^-7到10^-3都尝试过)
from douzero.
@daochenzha 牌堆的生成思路是先将牌的点数(3到A,2,小王,大王)打乱,再按照打乱后的顺序,王各生成1张,3到A和2各生成4张,再将生成的牌堆进行切牌,这样就能模仿不洗牌模式的“瑕疵”,而不至于“过于整齐”,整个过程只生成一次64位随机整数作为种子,牌的点数打乱方式、切牌次数以及切牌位置都通过种子取余来解决。
from douzero.
@daochenzha 按照洗牌模式进行训练,训练初期的loss在默认情况下只有个位数,但是不洗牌模式的loss可以在几十到几万之间波动。
from douzero.
如果用胜率进行训练则不会造成十分高的loss
from douzero.
如果用胜率进行训练则不会造成十分高的loss
受此启发,个人尝试修改了奖惩规则,减少了bomb_count对奖惩结果的影响,防止其指数扩大后数值偏差过于严重
from douzero.
from douzero.
这个新的奖惩机制鼓励AI在胜算很大的时候提高bomb count,同时大幅减少bomb count小于等于3时的惩罚幅度,这样AI不至于学会摆烂
from douzero.
Related Issues (20)
- 为什么同样是4个GPU我的训练时候的FPS很低呢,基本都在2000左右 HOT 7
- 关于models.py 几个问题 HOT 7
- 抢地主功能什么时候开放 HOT 1
- 求助,用CPU无法训练问题!!! HOT 1
- 策略调整可行性 HOT 2
- 执行loss.backward()几个小时报错out of memory HOT 7
- 策略可能性
- 咨询下LSTM策略调整
- 多显卡瓶颈在第0卡上
- Apple M1芯片支持模拟和训练吗 HOT 1
- colab上的训练代码一运行,ram就炸了 HOT 3
- 欢乐斗地主实测胜率很低 算法还有升级的办法吗? HOT 2
- 关于深度学习与强化学习 HOT 2
- 另外,训练过程中注意CPU和显存 我是actor用CPU,training用GPU,CPU一定要留余地,同时thread不要加太高 否则容易爆显存 HOT 1
- actor网络在收集数据时使用了model前向传播,leaner对网络参数再进行跟新,这两者对同一共享内存更新是否会有冲突? HOT 1
- 勾吧pretrain models都没有 一训练全是0 HOT 2
- 在colab在跑训练,时间到了可以从上一次继续么 HOT 1
- EOFError: Ran out of input HOT 2
- 想询问一下baseline里面的ai训练了多久,我训练了几个礼拜了,表现感觉不如baseline HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from douzero.