关于训练集设置的一些疑问 #31
-
最近尝试了一下把一些Mortal自我对抗的牌谱加入到训练集中,结果发现loss函数不正常地减少(直接减半),而且实战能力在统计数据上有很明显的退步(经过400万局训练后的5万场对局 avg_pt小于-1)。不清楚是不是设置上出了什么问题? |
Beta Was this translation helpful? Give feedback.
Answered by
Equim-chan
Nov 29, 2022
Replies: 2 comments 6 replies
-
如果训练集是当前模型自己产生的,那么 loss 大幅降低是很正常的。 |
Beta Was this translation helpful? Give feedback.
6 replies
Answer selected by
hyskylord
-
你好,可以分享下数据和训练的方法么,有偿 |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
如果训练集是当前模型自己产生的,那么 loss 大幅降低是很正常的。