Skip to content

Commit

Permalink
add readme,update image and elo
Browse files Browse the repository at this point in the history
  • Loading branch information
bupticybee committed Oct 9, 2018
1 parent bc2e8fe commit a331629
Show file tree
Hide file tree
Showing 5 changed files with 218 additions and 162 deletions.
4 changes: 2 additions & 2 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@

这个项目受到alpha go zero的启发,旨在训练一个中等人类水平或高于中等人类水平的深度神经网络,来完成下中国象棋的任务。目前这个项目仍在积极开发中,并且仍然没有完成全部的开发,欢迎pull request 或者star。 然而受到计算资源限制,这样庞大的任务不可能在一台机器上完成训练,这也是我完成了分布式训练代码的原因,希望各位小伙伴能够加入,一起训练这样一个中国象棋alpha go的网络。

我的估计是达到4000~5000elo分数的时候深度网络可以达到目标,现在深度网络已经到了1400分的边缘,达到人类中上水平的目标并不是不可能的,因为计算资源限制,我希望能用较少计算资源来进行整个训练,所以前几天我加入了一些人类棋谱,这也是elo曲线中跳跃的折线的原因
我的估计是达到4000~5000elo分数的时候深度网络可以达到目标,现在深度网络已经到了3000分的边缘,达到人类中上水平的目标并不是不可能的。

目前的elo:

Expand All @@ -12,7 +12,7 @@

![table](imgs/table.png)

当然,目前棋力还比较差,因为是从完全随机开始训练的,比方说某个对局片段:
当然,目前棋力还比较一般,因为是从完全随机开始训练的,比方说某个对局片段(800 playouts)

![demo](imgs/demo.gif)

Expand Down
Binary file modified imgs/elos.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file modified imgs/table.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
301 changes: 173 additions & 128 deletions ipynbs/elo_graph.ipynb

Large diffs are not rendered by default.

75 changes: 43 additions & 32 deletions ipynbs/see_gameplay.ipynb

Large diffs are not rendered by default.

0 comments on commit a331629

Please sign in to comment.