BLEU_4 score bug in LeaderBoard?

AIChallenger / AI_Challenger_2017

AI Challenger, a platform for open datasets and programming competitions to artificial intelligence (AI) talents around the world.

https://challenger.ai/

474 stars 196 forks source link

BLEU_4 score bug in LeaderBoard? #31

Open wangheda opened 6 years ago

wangheda commented 6 years ago

Our BLEU_4 score (0.73298) on the LB is approximately equal to the BLEU_1 score (0.73563) we measured offline on eval dataset, and much higher than our offline BLEU_4 score (0.41735).

I wonder if the score display on the LB is wrong.

chenghuige commented 6 years ago

@wangheda @AIChallenger 是这样的而且现在所有队伍的bleu@4值显示的应该都不是bleu@4的数值。。。