An easy whitening-based vector normalization strategy with less than 10 lines of code consistently boosts the performance of the model . The researchers say the embedding of a sentence in an unsupervised way is valuable tonatural language matching and retrieval problems in practice . They say combining both top andbottom layers is better than only using top layers . An easy-whiteening-by-white-language-normalization strategy is also a good way to whiten-based vectors .

Author(s) : Junjie Huang, Duyu Tang, Wanjun Zhong, Shuai Lu, Linjun Shou, Ming Gong, Daxin Jiang, Nan Duan

Links : PDF - Abstract

Code :


Keywords : easy - top - language - based - normalization -

Leave a Reply

Your email address will not be published. Required fields are marked *