GeTLeft
GeTLeft
One possible solution is to use a small lr for target encoder(e.g. 1e-5) and reduce training epoch. My setting: ResNet-50 for source and target encoder. lr for source encoder :...
BTW, I use https://github.com/corenel/pytorch-adda not this offical Tensorflow. My result (A->W) is 81% using above setting.
> Hi Tianlin-Gao, > > Thanks for your reply. May I know which layer are you matching and do you fix some layers in the source/target encoder to avoid overfitting?...
@githubharald I am installing CRNN on 16.04. In your comment, you said that using the correct versions for all libs is the key. I wonder that how to choose the...
@githubharald Thanks. have tried the install_all.sh and encountered plenty of errors. I am going to try the pytorch version.
Just a remainder. **Use 14.04 instead of 16.04.**
This error maybe because your ndarray type doesn't match c++ std::vector 's type. You can set ndarray type by `np.save(f, array.astype(np.float32))`
Try to lower the lr of target encoder and reduce the number of epoch. My setting: Dataset: Office 31 (A->W) Network: ResNet-50 for source and target encoder. Same discriminator as...
> 这是因为` hexo-renderer-marked` 渲染存在特殊字符转义和mathjax冲突导致的，可以采用下面方式解决。 > > 1. 在主题的`_config.yml`开启`mathjax` > 2. 在你要渲染的文章开头添加`mathjax: true` > 3. 在终端执行： > > ``` > npm uninstall hexo-renderer-marked --save > npm install hexo-renderer-kramed --save > ```...
现在最棘手的问题是zhihu登陆加载不出二维码，@niudai 有什么思路吗，我本职是写C++的，如果有思路可以试着修一下这个问题