Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Experimental models link broken #56

Open
erickrf opened this issue Feb 11, 2022 · 6 comments
Open

Experimental models link broken #56

erickrf opened this issue Feb 11, 2022 · 6 comments

Comments

@erickrf
Copy link

erickrf commented Feb 11, 2022

The link mentioned in the Readme to experimental models seems broken or with restricted access: https://github.com/Helsinki-NLP/Opus-MT-train/tree/master/work-spm/models

@erickrf erickrf changed the title Experimental models like broken Experimental models link broken Feb 11, 2022
@jorgtied
Copy link
Member

Good catch. I didn't update that list for a long time and decided to remove those sub-directories to make the repository a bit more easy to handle. Is there a specific model that you would need?

@erickrf
Copy link
Author

erickrf commented Feb 14, 2022

I was wondering if there was a German-Slovak model.

@jorgtied
Copy link
Member

I can check but I could also start a job to train one. Maybe a German --> Czech + Slovak would be a good choice?

@erickrf
Copy link
Author

erickrf commented Feb 14, 2022

Thanks, that would be great! I'm not an expert on these languages but I believe this combination makes sense.

@jorgtied
Copy link
Member

@erickrf: I have uploaded a model for de-sk now: https://object.pouta.csc.fi/OPUS-MT-models/de-sk/opus_transformer-align_2022-02-19.zip. The multilingual model with Czech is still running. More info about the de-sk model is here: https://object.pouta.csc.fi/OPUS-MT-models/de-sk/opus_transformer-align_2022-02-19.yml

@erickrf
Copy link
Author

erickrf commented Feb 21, 2022

Thanks! How can I create a huggingface transformers configuration from the checkpoint? Is it by the way much less efficient than using Marian directly?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants