[Adapt] [Seminar] Deep Contextualized Word Representations

[张盛瑶]

Hi Adapters：
   For seminar tomorrow, I will introduce a method of word embedding: ELMo. It is a deep contextualized word representation and word vectors are learned functions of the internal states of a deep bidirectional  language model. It has a simple architecture but when the representations added to existing models it can significantly improve the state-of-the-art across some challenging NLP tasks. You can find more details in this paper: https://arxiv.org/pdf/1802.05365.pdf
   See you then!


Time: 17:00 December 12
Venue: SEIEE 3-517A
Best Regards, 
Sophie

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ATT00003.txt
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20181211/5bf46435/attachment.txt>