Adding input layers to NMT models

[Jash Dubal]

Hello,

I've been working with Opus-Marian for the past few weeks, performing model re-training and fine-tuning, and the model performance is lacking.

We would like to explore the possibility of adding additional input layers into the training and inferencing process. For example, instead of sending an English sentence to translate, I would like to also provide metadata with the sentence (e.g. author name). I am wondering if anyone has tackled this and could provide some guidance on how to implement additional input layers and re-train a base model.

Any advice on how we can add additional inputs along with our translated texts would be extremely helpful.