How to fine-tune NLLB-200 model?

[epiniguin]

Hi!

How to train NLLB model using an existing NLLB-200 model(for example 3.3B) as a checkpoint?

[cbalioglu]

Hi @epiniguin, we are about to land a refactoring to our recipes that will reduce the amount of code one needs to write. Do you use a publicly available dataset? We plan to write READMEs for each recipe and we might use your use case as an example.

[epiniguin]

@cbalioglu I use publicly available datasets.

[epiniguin]

Hi @cbalioglu !

Are there any news about recipe refactoring and Readme?
I'm also interested in MT evaluation recipe.