I have a problem statement like Converting deprecated code into a modern version of the same language. I'm currently converting with a custom Rule-based engine. But the modern version of the language is so evolved that there are more than 100k modifications. So, I thought building an ML model for this purpose will solve the problem here.
Searching through the internet for any existing models for my problem statement, came across TransCoder.
My questions are :
- Can I use this model or Is there any other efficient way of solving this problem?
- Can I use the TransCoder repository code to train a model with my dataset or train the existing model to get high efficiency?
- In Transcoder, they have used up to 600GB of open source projects for training. Can I attain that efficiency with low dataset count? Or I have to alter the code to achieve that?
- Any suggestion to solve this problem, in general?
Thanks in advance.