word2vec_pytorch
word2vec_pytorch copied to clipboard
避免报GBK错误,修改了打开文件的方式 更新pytorch版本至1.1 关于小bug i==j 更改为 u==v(可能是作者吗字太快了)
def evaluate_pair_count(self, window_size): return self.sentence_length * (2 * window_size - 1) - ( self.sentence_count - 1) * (1 + window_size) * window_size where does this formula come from? Is it...
thans for your code, but when I trained the model, the loss can only be reduced to 130, I want to know if this is normal ?
您好,get_batch_pairs函数中的i==j是为了排除本身吗?但是这样的逻辑也排除不了本身啊,i==j的时候他们并不是同一个啊
** evaluate_pair_count()**这个方法得到的pair是4264098. 实际上在zhihu.txt中的pair应该是4746078.
can u update this for the new pytorch
如题,这样做有什么好处吗? 谢谢!