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? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
我想问下,是否可以用普通Python模块解析xdl.parsers.pb格式?xdl不太好编译。
The text was updated successfully, but these errors were encountered:
解析这个是为了什么呢?xdl 编译还好,装个docker 比较容啊
Sorry, something went wrong.
哦哦,好的,想另外问一些ESMM模型对应的Ali-CCP数据集的处理,有两个问题: 1)这个里面的所有特征都是放到一个embedding矩阵表去lookup的么?是否有测试过各自特征域分别建立embedding矩阵表的效果? 2)针对这个embedding矩阵表是进行了hash处理呢,还是做的raw千万级的embedding矩阵表进行训练呢?
没有使用过,不敢乱答。只说一下之前做的一些处理,会把特征映射到一个id空间内,比如gender-男映射为1 gender-女为2 ,这样基本一个embedding就可以表示所有特征id,xdl会把embedding 分散在多个ps 之间进行存储。 看到阿里天池的数据,你这是作比赛吗?
这个不是比赛数据,是阿里开源出来的多任务模型对应的数据集。
No branches or pull requests
我想问下,是否可以用普通Python模块解析xdl.parsers.pb格式?xdl不太好编译。
The text was updated successfully, but these errors were encountered: