issues
search
initial-h
/
AlphaZero_Gomoku_MPI
An asynchronous/parallel method of AlphaGo Zero algorithm with Gomoku
185
stars
43
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
C_puct 参数
#48
Sunshine-718
closed
2 weeks ago
1
把value head的输出改成像DQN一样的多头输出是否可行？
#47
Sunshine-718
closed
1 month ago
3
关于experience replay的问题
#46
Sunshine-718
closed
1 month ago
2
转换成pytorch
#45
ASL-zhenglu
closed
3 months ago
3
laptop for this project
#44
hharryyf
closed
1 year ago
1
Training time
#43
wiki-love
closed
1 year ago
3
reset the MCTS during self-play
#42
hharryyf
closed
1 year ago
2
Will the forbidden moves be considered in the future?
#41
iamlxb3
closed
1 year ago
1
关于训练局数以及网络大小的疑问
#40
ojbk888
closed
2 years ago
3
小白关于mpi多进程通信求助
#39
ojbk888
closed
2 years ago
2
感谢作者，这是我用过最有用的架构，我修改成了keras的模型，我有一个问题，当训练到后期的时候，episode_len会越来越小，在9-15左右
#38
yuan6785
closed
2 years ago
1
training for size of 19*19 and connect-six is so slow，what should I do？
#37
177715031
closed
2 years ago
1
how to modify code to AI VS AI?
#36
massawuyh
closed
3 years ago
2
I have a question about entropy
#35
1331han
closed
3 years ago
1
Why not use np.flipud to do dihedral reflection?
#34
leilani-x
closed
4 years ago
2
大神您好，有问题想请教一下
#33
161220028jcdai
closed
3 years ago
18
Hi, I'm the person who mailed 3 days before.
#32
ssm0414
closed
3 years ago
11
关于五子棋执黑第一步和禁手问题？
#31
LLYYKK
closed
4 years ago
2
关于白棋的退火温度的疑问
#30
sliver2000
closed
4 years ago
1
get_action()里,为什么作为标签的move_probs是tau始终为1得来的？
#29
apocalypse-revelation
closed
4 years ago
4
16G的内存，data_buffer设置多少合适呢？
#28
apocalypse-revelation
closed
4 years ago
11
有没有pytorch版本的呢？
#27
apocalypse-revelation
closed
4 years ago
4
关于模型的一些疑惑
#26
sliver2000
closed
4 years ago
5
关于MPI版本中棋力没有进步的问题？
#25
LLYYKK
closed
4 years ago
2
关于Dihedral Reflection or Rotation和node概率p的一些困惑
#24
luopeng69131
closed
4 years ago
4
关于MCTS_Pure的一些疑问？
#23
LLYYKK
closed
4 years ago
3
关于MPI版本的一些疑问？
#22
LLYYKK
closed
4 years ago
2
关于train.py的一些疑问
#21
LLYYKK
closed
4 years ago
4
作者你好，我想到了一个问题啊，之前的AiphaZeroGo的policy network和value network是分开的，而AiphaZero的两个网络合在一起了，这样做具体有什么好处呢
#20
baimantou
closed
4 years ago
1
Tricks Discussion
#19
SeriaQ
closed
3 years ago
10
作者你好，看了你的代碼，只有policy network 我很好奇你的value network，有沒有訓練
#18
baimantou
closed
4 years ago
35
训练15*15的棋盘
#17
Nkust-R105
closed
4 years ago
7
迁移学习很神奇啊，有意思！
#16
tangbh
closed
4 years ago
3
ubuntu make教程
#15
snail-ju
closed
4 years ago
1
請問這個專案有沒有使用到apv_mcts?
#14
will0206
closed
4 years ago
1
policy_value_net_tensorlayer.py一个小问题
#13
tangbh
closed
4 years ago
2
关于 τ 的疑问
#12
tangbh
closed
4 years ago
5
使用模型报错
#11
tangbh
closed
4 years ago
4
请问15x15 11x11 分别训练了多久啊？
#10
zxa308078684
closed
4 years ago
4
我用多进程训练的时候GPU和CPU都没有在使用
#9
REFunction
closed
4 years ago
3
关于MPI版的显卡利用率
#8
hijkzzz
closed
4 years ago
6
基于您提供的model再进行训练问题
#7
liunian321
closed
4 years ago
3
AI执黑不走天元
#6
Augustusmyc
closed
4 years ago
5
Initial h patch 1
#5
initial-h
closed
5 years ago
0
概率显示问题
#4
liunian321
closed
5 years ago
1
问一个比较低级的关于单GPU训练的问题
#3
bingshen
closed
5 years ago
2
有关该项目和原版的差异的一些疑问
#2
tml104
closed
5 years ago
2
基于小板训练大板上的AI疑问
#1
liunian321
closed
5 years ago
4