A multi-task learning approach (called Doduo), trainedusing pre-trained language models on both tasks, outperforms individual learning approaches . Doduo establishes new state-of-the-art performance on two benchmarks for the column type predictionand column relation prediction tasks with up to 4.0% and 11.9% improvements,respectively. We also establish that Doduo can already perform the previous performance with a minimal number of tokens, only 8 tokens per column . Dodio can already . perform the . previous performance . with a minimum number of . tokens, just 8 tokens, in order to perform .

Author(s) : Yoshihiko Suhara, Jinfeng Li, Yuliang Li, Dan Zhang, Çağatay Demiralp, Chen Chen, Wang-Chiew Tan

Links : PDF - Abstract

Code :
Coursera

Keywords : tokens - performance - column - perform - doduo -

Leave a Reply

Your email address will not be published.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