@rgcottrell @tianfeichen @cqlijingwei
I am actually researching about the different types of models for grammatical correction.
So, i used arch='transformer_iwslt_de_en' from this link,
https://github.com/pytorch/fairseq/tree/master/examples/translation#iwslt14-german-to-english-transformer
Here, i changed the model from fconv to transformer. I got really disappointing results for the test set, jfleg.
Do you want to share some thoughts? What newer features from fairseq do you think will be the fittest for the extension of your project?
Any help would be appreciated. Thanks in advance.
@rgcottrell @tianfeichen @cqlijingwei
I am actually researching about the different types of models for grammatical correction.
So, i used arch='transformer_iwslt_de_en' from this link,
https://github.com/pytorch/fairseq/tree/master/examples/translation#iwslt14-german-to-english-transformer
Here, i changed the model from fconv to transformer. I got really disappointing results for the test set, jfleg.
Do you want to share some thoughts? What newer features from fairseq do you think will be the fittest for the extension of your project?
Any help would be appreciated. Thanks in advance.