Skip to content

Add support for M2M100EncoderModel (aka NLLB, aka SONAR text encoder)#2541

Open
avidale wants to merge 1 commit into
huggingface:mainfrom
avidale:sonar-encoders
Open

Add support for M2M100EncoderModel (aka NLLB, aka SONAR text encoder)#2541
avidale wants to merge 1 commit into
huggingface:mainfrom
avidale:sonar-encoders

Conversation

@avidale
Copy link
Copy Markdown

@avidale avidale commented Mar 13, 2024

This is a draft pull request for supporting SONAR text encoder model.
SONAR encoder is architecturally an encoder from NLLB, which is, in turn, is implemented as M2M100 in HF Transformers.
Thus, I support the encoder-only part of it, in the same way as T5-based encoders are currently supported.

This PR depends on a pull request to Transformers huggingface/transformers#29646, and should be reviewed only after that one.

Probably, the minimal required version of Transformers should also be updated to point to a release after the aforementioned PR is merged.

@tomaarsen
Copy link
Copy Markdown
Member

Hello!

I'll be glad to merge this if Transformers accepts your work. Feel free to ping me when that happens.

  • Tom Aarsen

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

Successfully merging this pull request may close these issues.

2 participants