demo quick_start中dataprovider_bow.py特征对齐问题
Created by: donote
fixes https://github.com/PaddlePaddle/Paddle/issues/676
该文件采用BOW将comments转换为特征向量,在settings中sparse_binary_vector为3w维。而在process中word_vector是不定长的,请问word_vector作为输出数据,如何处理特征对齐的问题,word_vector和settings.sparse_binary_vector有什么关系呢,谢谢。代码如下:
sparse_binary_vector(len(dictionary))
word_vector = [settings.word_dict.get(w, UNK_IDX) for w in words]