Skip to content

27-th place solution for Feedback Prize - English Language Learning

Notifications You must be signed in to change notification settings

andrei-papou/kaggle-fp-ell

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

44 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

FP-ELL Competition Solution

27-th place solution for Feedback Prize - English Language Learning

Most of the notebooks use kaggle-toolbox library under the hood.

Short summary

The key idea of the solution is to use single target models instead of the multi-target one. While generally fitting a multi-target model is a good idea, experiments clearly show that the single-target ones perform significatly better. Possible explanation is a high correlation between the targets combined with a huge difference in the task complexity.

About

27-th place solution for Feedback Prize - English Language Learning

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published