search

xing-hu / EMSE-DeepCom

The dataset for EMSE-DeepCom
MIT License
118 stars 63 forks source link

Why the corpus_bleu is 0.0000? #13

Open 9ayhub opened 4 years ago

9ayhub commented 4 years ago

I use data-v1, the data set you provide, to train this model. It looks strange that the results of corpus_bleu are always 0.0000, did I do something wrong ？

07/11 02:15:34 test score=0.0000 avg_score=0.3086
07/11 03:33:38 test score=0.0000 avg_score=0.3993
07/11 04:51:08 test score=0.0000 avg_score=0.3668
codekali commented 4 years ago

Did you train the model before calculating the bleu score?

satinewee commented 3 years ago

I use data-v1, the data set you provide, to train this model. It looks strange that the results of corpus_bleu are always 0.0000, did I do something wrong ？

07/11 02:15:34 test score=0.0000 avg_score=0.3086
07/11 03:33:38 test score=0.0000 avg_score=0.3993
07/11 04:51:08 test score=0.0000 avg_score=0.3668

Now I meet the same problem sa yours, and I have trained the model.How did you solve this problem?