Skip to Main Content
Table A.2 

Statistics of data used for tuning. The numbers of target tokens are averages across four reference translations for ZH→EN and UR→EN, rounded to the nearest token.


lines
source tokens
target tokens
ZH→EN 919 24,152 28,870 
DE→EN 1,300 29,791 31,318 
UR→EN 882 18,004 16,606 
EN→MG 1,359 28,408 32,682 

lines
source tokens
target tokens
ZH→EN 919 24,152 28,870 
DE→EN 1,300 29,791 31,318 
UR→EN 882 18,004 16,606 
EN→MG 1,359 28,408 32,682 
Close Modal

or Create an Account

Close Modal
Close Modal