self.embedding = nn.Embedding(self.vocab_size + 2, self.embedding_dim, padding_idx=self.vocab_size + 1)
self.embedding.weight = nn.Parameter(opt.embeddings)
pytorch0.3.1时这样设置embedding，没有问题
但0.4.1版本nn.embedding函数有改变，可以直接将embeddings作为参数，这样调用：
self.embedding = nn.Embedding(self.vocab_size, self.embedding_dim, padding_idx=self.vocab_size-1, _weight=opt.embeddings)
可以看做将vocab的最后一个词用zero向量padding
pytorch0.3.1时这样设置embedding，没有问题但0.4.1版本nn.embedding函数有改变，...
安卓报错error: uncompiled PNG file passed as argument. Must b...
安卓报错error: uncompiled PNG file passed as argument. Must b...
按照官方文件 “csvfile如果是文件则must be opened with the ‘b’.” 报错 _cs...
去掉？报错为： Value of optional type '[MKAnnotation]?' must be ...
运行报错： does not contain bitcode. You must rebuild it with ...
conda list 报错 TypeError: LoadLibrary() argument 1 must be...
报错如下: StatefulWidget.createState must return a subtype of...
报错：This request has been blocked; the content must be ser...
报错1：The "path" argument must be of type string. Received ...
本文标题：pytorch0.4.1报错Padding_idx must b
本文链接：https://www.haomeiwen.com/subject/dlyobftx.html
网友评论