Finetuning a model for machine translation on a programming language

Hi all,

I have a neural machine translation problem, I have my own custom dataset and my dataset is not natural language. Both source and target languages are code.

Can I use all models that have a ConditionalGeneration head for this task? Most of these models mention summarization task but they do not say machine translation. Also, on the models page where one can search for all uploaded models, I only see T5 and FMST if I filter out the search with “translation”.

Could you clarify the situation in general and which models would you suggest?

We too are searching models for source-to-source conversion, aka, transpiling.
What models have you found that perform this task?