Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

非Bert验证F1始终保持不变以及中文文本loss上升验证F1才上升的问题 #212

Open
Ya-dongLi opened this issue Aug 21, 2022 · 0 comments

Comments

@Ya-dongLi
Copy link

以下问题是使用非bert的模型出现的:
1.对于您原本的代码,我们用的词向量字典为glove.42B.300d,restaurant数据集,跑了30轮,除了一开始的两轮验证F1为10%,之后的28轮的验证F1始终保持26.26%不变,尚不清楚是哪个环节出了问题(绝大部分模型都是这样)。
38d6095312baf030aaa90fb66a4af43
35239b7349e317b50360fb5fe2eb3b7
c09da5f0c41098dec24ec4fd545f18e
(每轮结果都是这样,这里只展示几张)
2.之后我们更换为中文文本,词向量采用了Chinese-Word-Vectors,数据集是自己标注的经济类数据,采用jieba分词。对于模型名字带lstm的模型,也是上述问题,验证F1始终保持不变,对于cabasc和memnet则会有较大波动,最佳会有0.75的F1,但是出现的问题是,当训练loss很低的时候(1左右),F1很低,当loss很高的时候(几千),F1反而上去了。而对于bert就没有类似问题。
abf0fe9d7ceeaddda0e0913c6c3e6f2
91f9466aa589204dd24f8055039b6cd

我尝试修改过学习率,数据量,但是结果并不会有太大改变,最多会有稍微的一点点波动(1%-2%)。目前找不到问题的原因,十分苦恼。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant