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
NER代码运行问题 #15
Comments
Hi, I have uploaded the checkpoint file for Weibo NER, in which there is a shell file to train and evaluate the model. I suggest you train the model using that shell file. Hopes it help. Wei |
Hi, How about the result loading my checkpoints? |
加载您给的checkpoint在验证集和测试集上的结果是正常的,f1跟论文中的值一样 |
Hi, Please be patient. I will try to run the code to find if there are bugs in it. |
thank you very much! |
我遇到过类似问题,同样的代码在1080Ti-11g显卡和V100-16G显卡上结果完全不一样,在V100显卡上一直不收敛,f1在前几十个epoch上一直为0。可以尝试换一台机器。 |
多谢提醒,我在别的卡上试试~ |
果然,是cuda版本的问题,用cuda10.1运行loss下降正常 |
@hezongfeng Thanks for your sharing. I don't know the version of CUDA and machine will make such a difference. @JucksonP Sorry to reply so late. My running status is normal, and the loss has dropped. I am a full-time employee in the company so don't have so much time to run the experiment. Please understand. |
And my GPU is tesla-P100. But I am not sure about the version of Cuda since it may be changed by my colleagues. Now, the version of two GPUs I used in the last year is 10.2 and 11.0 |
单卡训练下,遇到过类似的问题,替换CRF层解决。 |
你好,我的邮箱号是1162276945@qq.com |
你好,请问你是用什么替换CRF层的呢? |
我晕,这么奢侈呀,40G的显卡,我在自己电脑上8G显卡RTX3070跑不动,我猜想可能是这个词库太大了。导致Trie树异常庞大。 |
公司服务器就是香,24张A100哈哈哈哈,显卡我是买不起的 |
你好,大佬:
首先,感谢开源!
尝试复现论文结果的时候遇到了一些问题,不知能否抽空解答一下.
1、在运行weiboNER的实验代码时,超参数设置与论文中一样,训练时loss下降有些异常(震荡下降,且前几个epoch验证集测试集f1均为0),训练日志已邮件发送;
2、具体环境及运行设置:
GPU:A100-SXM4-40GB; torch:1.8.1+cu111 训练方式:单卡
期待大佬回复指导,谢谢!
The text was updated successfully, but these errors were encountered: