BLEU score 계산하기 #17

eubinecto · 2022-06-02T15:19:55Z

Why?

#16에서 학습을 마무리한 모델의 성능이 얼마나 좋은지 파악을 해야한다. 그러기 위해선 믿을만한 지표를 하나 선정해서
정량적인 평가를 진행해야한다. 그리고 번역에서 믿을만한 지표로는 BLEU가 있다.

eubinecto · 2022-06-03T04:47:08Z

d

eubinecto mentioned this issue Jun 2, 2022

Chronicles #14

Open

9 tasks

eubinecto added a commit that referenced this issue Jun 2, 2022

[#17] main_eval.py added

65e6d93

eubinecto added a commit that referenced this issue Jun 2, 2022

[#17] main_eval.py added

faf0c12

eubinecto added a commit that referenced this issue Jun 2, 2022

[#17] main_eval.py first attempt at fixing the bug

0f6cda7

eubinecto added a commit that referenced this issue Jun 2, 2022

[#17] registering tokenizer externally

f7ebe0c

eubinecto added a commit that referenced this issue Jun 3, 2022

[#17] callbacks implemented

966f206

eubinecto added a commit that referenced this issue Jun 3, 2022

[#17] no longer tagging tokenizer with bpe or wp

cc6a87d

eubinecto added a commit that referenced this issue Jun 3, 2022

[#17] tagging runs with __file__

06e0394

eubinecto mentioned this issue Jun 3, 2022

Issue 17 #18

Merged

eubinecto closed this as completed Jun 3, 2022

eubinecto changed the title ~~main_eval.py (bleu score 계산하기)~~ BLEU score 계산하기 Jun 3, 2022

eubinecto added a commit that referenced this issue Jun 3, 2022

[#17] main_build_kor2eng.py done. Makefile has been added

f871889