- ๐ฌ Iโm a researcher in Microsoft Reserach Asia.
- ๐ญ Iโm currently working on the pre-trained models for language and vision (Language: UniLMv2 & Vision: BEiT & Langugae + Vision: BEiT-3.
- ๐ซ How to reach me: email: x@y, x = addf400 & y = foxmail.com. Google scholar profile.
addf400 / tupe Goto Github PK
View Code? Open in Web Editor NEWThis project forked from guolinke/tupe
Transformer with Untied Positional Encoding (TUPE). Code of paper "Rethinking the Positional Encoding in Language Pre-training".
License: MIT License