Skip to content

Contributing trained Greek<->English NMT models implemented with fairseq #8233

Description

@lighteternal

Hi there, quick question that I couldn't answer by searching the docs:

I trained an EL-EN (Greek to English) and an EN-EL machine translation model using the fairseq implementation of the transformer_iwslt_de_en architecture on ~6GB of parallel corpora. Given that the models report a better BLEU score compared to the existing SotA, I would like to share them somehow. I thought that fairseq might offer a huggingface-like way to upload trained models but I couldn't find any, so I would appreciate any guidance.

If there's a straightforward way to convert and upload these as huggingface models it would be great!

Many thanks!

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions