We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
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
想训练压缩版的FOM模型,根据doc上的教程,先训练了kp_detector,最终收敛结果如下: 可以看到同比损失在0.2左右 然后继续训练generator,最终收敛结果如下: generator 训练的效果不好,但也可以接受 最终两个一起训练,修改了配置中的mode为both,以及两个参数路径为上面两个路径里的checkpoint.pdparams,如下: 这个和刚开始就采用both的mode方式训练没有任何区别,两个权重的参数根本没用上
使用的是develop分之,我看作者把里面部分7X7的卷积核拆成了3个3X3的小卷积,这个我之前在pytorch的模型里尝试过,不过这个应该不影响。
训练的配置文件使用的是firstorder_vox_mobile_256.yaml ,数据集用的voxceleb1的经过预处理的,配置除了mode和两个权重参数路径,其它均没有改动,能问下这个现象正常吗,还是需要调整什么配置
The text was updated successfully, but these errors were encountered:
您好,请问问题还是否需要解决,目前相关图像生成能力集成在PaddleMIX中,https://github.com/PaddlePaddle/PaddleMIX/tree/develop 可以在这个repo下提出您的需求
Sorry, something went wrong.
lzzyzlbb
jerrywgz
No branches or pull requests
想训练压缩版的FOM模型,根据doc上的教程,先训练了kp_detector,最终收敛结果如下:
![image](https://user-images.githubusercontent.com/15072820/142101444-7471035b-fbbc-47b1-9dbf-47159917a4ee.png)
![image](https://user-images.githubusercontent.com/15072820/142102460-206963e5-78f6-44c6-9efd-4fa24e8a23ec.png)
![image](https://user-images.githubusercontent.com/15072820/142103598-3e906090-c14c-411c-9df9-03568b32f180.png)
可以看到同比损失在0.2左右
然后继续训练generator,最终收敛结果如下:
generator 训练的效果不好,但也可以接受
最终两个一起训练,修改了配置中的mode为both,以及两个参数路径为上面两个路径里的checkpoint.pdparams,如下:
这个和刚开始就采用both的mode方式训练没有任何区别,两个权重的参数根本没用上
使用的是develop分之,我看作者把里面部分7X7的卷积核拆成了3个3X3的小卷积,这个我之前在pytorch的模型里尝试过,不过这个应该不影响。
训练的配置文件使用的是firstorder_vox_mobile_256.yaml ,数据集用的voxceleb1的经过预处理的,配置除了mode和两个权重参数路径,其它均没有改动,能问下这个现象正常吗,还是需要调整什么配置
The text was updated successfully, but these errors were encountered: